koolamusic / flux-dev-lora-mahi
- Public
- 243 runs
-
H100
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDjvvpp2v4r9rmc0cp98kszpf5srStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.3
- output_quality
- 80
- prompt_strength
- 0.5
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "image": "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.3, output_quality: 80, prompt_strength: 0.5, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.3' \ -i 'output_quality=80' \ -i 'prompt_strength=0.5' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T18:27:04.820377Z", "created_at": "2025-04-18T18:26:57.858000Z", "data_removed": false, "error": null, "id": "jvvpp2v4r9rmc0cp98kszpf5sr", "input": { "image": "https://replicate.delivery/pbxt/MrMXqRPxd2vmMKPc7aa2Ddswje29nKDV6tZj6yOJGWj8aHVr/WhatsApp%20Image%202025-04-17%20at%2016.25.59.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 47474\nPrompt: Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:04, 3.43it/s]\n 13%|█▎ | 2/15 [00:00<00:04, 3.09it/s]\n 20%|██ | 3/15 [00:00<00:04, 2.97it/s]\n 27%|██▋ | 4/15 [00:01<00:03, 2.94it/s]\n 33%|███▎ | 5/15 [00:01<00:03, 2.91it/s]\n 40%|████ | 6/15 [00:02<00:03, 2.89it/s]\n 47%|████▋ | 7/15 [00:02<00:02, 2.88it/s]\n 53%|█████▎ | 8/15 [00:02<00:02, 2.88it/s]\n 60%|██████ | 9/15 [00:03<00:02, 2.87it/s]\n 67%|██████▋ | 10/15 [00:03<00:01, 2.87it/s]\n 73%|███████▎ | 11/15 [00:03<00:01, 2.87it/s]\n 80%|████████ | 12/15 [00:04<00:01, 2.86it/s]\n 87%|████████▋ | 13/15 [00:04<00:00, 2.86it/s]\n 93%|█████████▎| 14/15 [00:04<00:00, 2.86it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.86it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.89it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 6.3253313030000005, "total_time": 6.962377 }, "output": [ "https://replicate.delivery/xezq/3BfweC7fxjFZPo7Weycb8lzRyoTISvwssDTX63AndfRFvlgkC/out-0.webp" ], "started_at": "2025-04-18T18:26:58.495046Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-pog7hpv3fhdslohkxzi74plftwhr66bdy2owvthk3ifg3cjwmrwq", "get": "https://api.replicate.com/v1/predictions/jvvpp2v4r9rmc0cp98kszpf5sr", "cancel": "https://api.replicate.com/v1/predictions/jvvpp2v4r9rmc0cp98kszpf5sr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 47474 Prompt: Generate an ultra-detailed full-body portrait of <TOKMAHI>, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:04, 3.43it/s] 13%|█▎ | 2/15 [00:00<00:04, 3.09it/s] 20%|██ | 3/15 [00:00<00:04, 2.97it/s] 27%|██▋ | 4/15 [00:01<00:03, 2.94it/s] 33%|███▎ | 5/15 [00:01<00:03, 2.91it/s] 40%|████ | 6/15 [00:02<00:03, 2.89it/s] 47%|████▋ | 7/15 [00:02<00:02, 2.88it/s] 53%|█████▎ | 8/15 [00:02<00:02, 2.88it/s] 60%|██████ | 9/15 [00:03<00:02, 2.87it/s] 67%|██████▋ | 10/15 [00:03<00:01, 2.87it/s] 73%|███████▎ | 11/15 [00:03<00:01, 2.87it/s] 80%|████████ | 12/15 [00:04<00:01, 2.86it/s] 87%|████████▋ | 13/15 [00:04<00:00, 2.86it/s] 93%|█████████▎| 14/15 [00:04<00:00, 2.86it/s] 100%|██████████| 15/15 [00:05<00:00, 2.86it/s] 100%|██████████| 15/15 [00:05<00:00, 2.89it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDfbq35ez2qxrmc0cp99y88hfb7cStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- seed
- 343411
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard.
- go_fast
- lora_scale
- 1.133
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.111
- output_quality
- 80
- prompt_strength
- 0.3784
- extra_lora_scale
- 1
- num_inference_steps
- 33
{ "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.133, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.111, "output_quality": 80, "prompt_strength": 0.3784, "extra_lora_scale": 1, "num_inference_steps": 33 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { seed: 343411, image: "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", go_fast: false, lora_scale: 1.133, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.111, output_quality: 80, prompt_strength: 0.3784, extra_lora_scale: 1, num_inference_steps: 33 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": False, "lora_scale": 1.133, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.111, "output_quality": 80, "prompt_strength": 0.3784, "extra_lora_scale": 1, "num_inference_steps": 33 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.133, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.111, "output_quality": 80, "prompt_strength": 0.3784, "extra_lora_scale": 1, "num_inference_steps": 33 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'seed=343411' \ -i 'image="https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard."' \ -i 'go_fast=false' \ -i 'lora_scale=1.133' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.111' \ -i 'output_quality=80' \ -i 'prompt_strength=0.3784' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=33'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.133, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.111, "output_quality": 80, "prompt_strength": 0.3784, "extra_lora_scale": 1, "num_inference_steps": 33 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T20:00:30.245030Z", "created_at": "2025-04-18T20:00:20.671000Z", "data_removed": false, "error": null, "id": "fbq35ez2qxrmc0cp99y88hfb7c", "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.133, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.111, "output_quality": 80, "prompt_strength": 0.3784, "extra_lora_scale": 1, "num_inference_steps": 33 }, "logs": "Loaded LoRAs in 2.68s\nUsing seed: 343411\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] img2img mode\n 0%| | 0/13 [00:00<?, ?it/s]\n 8%|▊ | 1/13 [00:00<00:03, 3.42it/s]\n 15%|█▌ | 2/13 [00:00<00:03, 3.10it/s]\n 23%|██▎ | 3/13 [00:00<00:03, 3.01it/s]\n 31%|███ | 4/13 [00:01<00:03, 2.97it/s]\n 38%|███▊ | 5/13 [00:01<00:02, 2.94it/s]\n 46%|████▌ | 6/13 [00:02<00:02, 2.92it/s]\n 54%|█████▍ | 7/13 [00:02<00:02, 2.91it/s]\n 62%|██████▏ | 8/13 [00:02<00:01, 2.91it/s]\n 69%|██████▉ | 9/13 [00:03<00:01, 2.91it/s]\n 77%|███████▋ | 10/13 [00:03<00:01, 2.91it/s]\n 85%|████████▍ | 11/13 [00:03<00:00, 2.92it/s]\n 92%|█████████▏| 12/13 [00:04<00:00, 2.91it/s]\n100%|██████████| 13/13 [00:04<00:00, 2.91it/s]\n100%|██████████| 13/13 [00:04<00:00, 2.94it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 7.759712139, "total_time": 9.57403 }, "output": [ "https://replicate.delivery/xezq/netHhyrscp2NdK2Wo8herrPCxxneez6arwnH4SiHi2M4VYQSB/out-0.webp" ], "started_at": "2025-04-18T20:00:22.485318Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-loympgozlfu3kxroi3qquy44lftlgd5zwnpusdwrex3l5cc6mv3a", "get": "https://api.replicate.com/v1/predictions/fbq35ez2qxrmc0cp99y88hfb7c", "cancel": "https://api.replicate.com/v1/predictions/fbq35ez2qxrmc0cp99y88hfb7c/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 2.68s Using seed: 343411 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard. Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] img2img mode 0%| | 0/13 [00:00<?, ?it/s] 8%|▊ | 1/13 [00:00<00:03, 3.42it/s] 15%|█▌ | 2/13 [00:00<00:03, 3.10it/s] 23%|██▎ | 3/13 [00:00<00:03, 3.01it/s] 31%|███ | 4/13 [00:01<00:03, 2.97it/s] 38%|███▊ | 5/13 [00:01<00:02, 2.94it/s] 46%|████▌ | 6/13 [00:02<00:02, 2.92it/s] 54%|█████▍ | 7/13 [00:02<00:02, 2.91it/s] 62%|██████▏ | 8/13 [00:02<00:01, 2.91it/s] 69%|██████▉ | 9/13 [00:03<00:01, 2.91it/s] 77%|███████▋ | 10/13 [00:03<00:01, 2.91it/s] 85%|████████▍ | 11/13 [00:03<00:00, 2.92it/s] 92%|█████████▏| 12/13 [00:04<00:00, 2.91it/s] 100%|██████████| 13/13 [00:04<00:00, 2.91it/s] 100%|██████████| 13/13 [00:04<00:00, 2.94it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDnq2hrzm281rma0cp9a09bdpr1cStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- seed
- 343411
- model
- schnell
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard.
- go_fast
- lora_scale
- 1.19
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.16
- output_quality
- 80
- prompt_strength
- 0.36
- extra_lora_scale
- 1
- num_inference_steps
- 12
{ "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "schnell", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.19, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.16, "output_quality": 80, "prompt_strength": 0.36, "extra_lora_scale": 1, "num_inference_steps": 12 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { seed: 343411, image: "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", model: "schnell", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", go_fast: false, lora_scale: 1.19, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.16, output_quality: 80, prompt_strength: 0.36, extra_lora_scale: 1, num_inference_steps: 12 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "schnell", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": False, "lora_scale": 1.19, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.16, "output_quality": 80, "prompt_strength": 0.36, "extra_lora_scale": 1, "num_inference_steps": 12 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "schnell", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.19, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.16, "output_quality": 80, "prompt_strength": 0.36, "extra_lora_scale": 1, "num_inference_steps": 12 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'seed=343411' \ -i 'image="https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png"' \ -i 'model="schnell"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard."' \ -i 'go_fast=false' \ -i 'lora_scale=1.19' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.16' \ -i 'output_quality=80' \ -i 'prompt_strength=0.36' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=12'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "schnell", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.19, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.16, "output_quality": 80, "prompt_strength": 0.36, "extra_lora_scale": 1, "num_inference_steps": 12 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T20:04:55.499998Z", "created_at": "2025-04-18T20:04:18.112000Z", "data_removed": false, "error": null, "id": "nq2hrzm281rma0cp9a09bdpr1c", "input": { "seed": 343411, "image": "https://replicate.delivery/pbxt/MrNyPYDX6om2UpxJOxpMw1CjawpG8yCg38ie7ZgIAnnoW0JD/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2007_48_56%20PM.png", "model": "schnell", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1.19, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.16, "output_quality": 80, "prompt_strength": 0.36, "extra_lora_scale": 1, "num_inference_steps": 12 }, "logs": "Loaded LoRAs in 0.55s\nUsing seed: 343411\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] img2img mode\n 0%| | 0/5 [00:00<?, ?it/s]\n 20%|██ | 1/5 [00:00<00:01, 3.58it/s]\n 40%|████ | 2/5 [00:00<00:00, 3.24it/s]\n 60%|██████ | 3/5 [00:00<00:00, 3.13it/s]\n 80%|████████ | 4/5 [00:01<00:00, 3.09it/s]\n100%|██████████| 5/5 [00:01<00:00, 3.07it/s]\n100%|██████████| 5/5 [00:01<00:00, 3.13it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 2.840926863, "total_time": 37.387998 }, "output": [ "https://replicate.delivery/xezq/VGEbLJLOfB18FyKeubGUo4W8ysd0XCUAyId2nRzSpjSnJGkUA/out-0.webp" ], "started_at": "2025-04-18T20:04:52.659071Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-7ozvcmsejbz5qg6guvwvpv67er4brjuckhqhbsxc4u7yhqyhye6a", "get": "https://api.replicate.com/v1/predictions/nq2hrzm281rma0cp9a09bdpr1c", "cancel": "https://api.replicate.com/v1/predictions/nq2hrzm281rma0cp9a09bdpr1c/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.55s Using seed: 343411 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard. Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] img2img mode 0%| | 0/5 [00:00<?, ?it/s] 20%|██ | 1/5 [00:00<00:01, 3.58it/s] 40%|████ | 2/5 [00:00<00:00, 3.24it/s] 60%|██████ | 3/5 [00:00<00:00, 3.13it/s] 80%|████████ | 4/5 [00:01<00:00, 3.09it/s] 100%|██████████| 5/5 [00:01<00:00, 3.07it/s] 100%|██████████| 5/5 [00:01<00:00, 3.13it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDner060ykxhrmc0cp9aza8hdyk0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.3
- extra_lora_scale
- 1
- num_inference_steps
- 37
{ "image": "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.3, extra_lora_scale: 1, num_inference_steps: 37 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.3' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=37'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T21:12:26.196062Z", "created_at": "2025-04-18T21:12:22.252000Z", "data_removed": false, "error": null, "id": "ner060ykxhrmc0cp9aza8hdyk0", "input": { "image": "https://replicate.delivery/pbxt/MrP4fE4nFvirDZ773KT12SMPG47qCkzrltL7NN3ABBthYzNi/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 58207\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] img2img mode\n 0%| | 0/12 [00:00<?, ?it/s]\n 8%|▊ | 1/12 [00:00<00:01, 5.59it/s]\n 17%|█▋ | 2/12 [00:00<00:01, 5.18it/s]\n 25%|██▌ | 3/12 [00:00<00:01, 5.12it/s]\n 33%|███▎ | 4/12 [00:00<00:01, 5.06it/s]\n 42%|████▏ | 5/12 [00:00<00:01, 5.03it/s]\n 50%|█████ | 6/12 [00:01<00:01, 5.01it/s]\n 58%|█████▊ | 7/12 [00:01<00:01, 4.99it/s]\n 67%|██████▋ | 8/12 [00:01<00:00, 5.00it/s]\n 75%|███████▌ | 9/12 [00:01<00:00, 5.00it/s]\n 83%|████████▎ | 10/12 [00:01<00:00, 4.98it/s]\n 92%|█████████▏| 11/12 [00:02<00:00, 4.99it/s]\n100%|██████████| 12/12 [00:02<00:00, 4.98it/s]\n100%|██████████| 12/12 [00:02<00:00, 5.03it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 3.930416925, "total_time": 3.944062 }, "output": [ "https://replicate.delivery/xezq/YBIsVevSXZQAASG43ZL8qGkmerK9kjG8tupfNqaeBOuqjcQSB/out-0.webp" ], "started_at": "2025-04-18T21:12:22.265645Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-wjphdw742fxl6cegtqlg3glmcrcrikzcxofmfguucilmzd32avfa", "get": "https://api.replicate.com/v1/predictions/ner060ykxhrmc0cp9aza8hdyk0", "cancel": "https://api.replicate.com/v1/predictions/ner060ykxhrmc0cp9aza8hdyk0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 58207 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] img2img mode 0%| | 0/12 [00:00<?, ?it/s] 8%|▊ | 1/12 [00:00<00:01, 5.59it/s] 17%|█▋ | 2/12 [00:00<00:01, 5.18it/s] 25%|██▌ | 3/12 [00:00<00:01, 5.12it/s] 33%|███▎ | 4/12 [00:00<00:01, 5.06it/s] 42%|████▏ | 5/12 [00:00<00:01, 5.03it/s] 50%|█████ | 6/12 [00:01<00:01, 5.01it/s] 58%|█████▊ | 7/12 [00:01<00:01, 4.99it/s] 67%|██████▋ | 8/12 [00:01<00:00, 5.00it/s] 75%|███████▌ | 9/12 [00:01<00:00, 5.00it/s] 83%|████████▎ | 10/12 [00:01<00:00, 4.98it/s] 92%|█████████▏| 11/12 [00:02<00:00, 4.99it/s] 100%|██████████| 12/12 [00:02<00:00, 4.98it/s] 100%|██████████| 12/12 [00:02<00:00, 5.03it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDf73n336hw5rm80cp9b0bnmkejrStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard
- go_fast
- lora_scale
- 1.11
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 0.3
- extra_lora_scale
- 1
- num_inference_steps
- 37
{ "image": "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.11, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", go_fast: false, lora_scale: 1.11, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 0.3, extra_lora_scale: 1, num_inference_steps: 37 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": False, "lora_scale": 1.11, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.11, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard"' \ -i 'go_fast=false' \ -i 'lora_scale=1.11' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=0.3' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=37'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\\nMaintain <TOKMAHI>\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.11, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T21:14:37.663745Z", "created_at": "2025-04-18T21:14:32.801000Z", "data_removed": false, "error": null, "id": "f73n336hw5rm80cp9b0bnmkejr", "input": { "image": "https://replicate.delivery/pbxt/MrP6jRVekjErFHp1eAp0iOlrSIL9zw4EGzFTgIHrgxtgIjUn/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard", "go_fast": false, "lora_scale": 1.11, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 37 }, "logs": "Loaded LoRAs in 0.67s\nUsing seed: 27514\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image.\nMaintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] img2img mode\n 0%| | 0/12 [00:00<?, ?it/s]\n 8%|▊ | 1/12 [00:00<00:01, 5.50it/s]\n 17%|█▋ | 2/12 [00:00<00:01, 5.16it/s]\n 25%|██▌ | 3/12 [00:00<00:01, 5.06it/s]\n 33%|███▎ | 4/12 [00:00<00:01, 5.02it/s]\n 42%|████▏ | 5/12 [00:00<00:01, 5.00it/s]\n 50%|█████ | 6/12 [00:01<00:01, 4.98it/s]\n 58%|█████▊ | 7/12 [00:01<00:01, 4.95it/s]\n 67%|██████▋ | 8/12 [00:01<00:00, 4.92it/s]\n 75%|███████▌ | 9/12 [00:01<00:00, 4.94it/s]\n 83%|████████▎ | 10/12 [00:02<00:00, 4.95it/s]\n 92%|█████████▏| 11/12 [00:02<00:00, 4.95it/s]\n100%|██████████| 12/12 [00:02<00:00, 4.95it/s]\n100%|██████████| 12/12 [00:02<00:00, 4.98it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 4.015968108, "total_time": 4.862745 }, "output": [ "https://replicate.delivery/xezq/F0JoP5vkPuJ6JpD9yek9GoGlfwKkImQA17WEEVtQKPP9KHkUA/out-0.webp" ], "started_at": "2025-04-18T21:14:33.647777Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-ddgnjfjysgxubcmuum7aozaacqueahxvkj7k4khbqselqxx5yopa", "get": "https://api.replicate.com/v1/predictions/f73n336hw5rm80cp9b0bnmkejr", "cancel": "https://api.replicate.com/v1/predictions/f73n336hw5rm80cp9b0bnmkejr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.67s Using seed: 27514 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. Maintain <TOKMAHI>'s facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit <TOKMAHI> exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] img2img mode 0%| | 0/12 [00:00<?, ?it/s] 8%|▊ | 1/12 [00:00<00:01, 5.50it/s] 17%|█▋ | 2/12 [00:00<00:01, 5.16it/s] 25%|██▌ | 3/12 [00:00<00:01, 5.06it/s] 33%|███▎ | 4/12 [00:00<00:01, 5.02it/s] 42%|████▏ | 5/12 [00:00<00:01, 5.00it/s] 50%|█████ | 6/12 [00:01<00:01, 4.98it/s] 58%|█████▊ | 7/12 [00:01<00:01, 4.95it/s] 67%|██████▋ | 8/12 [00:01<00:00, 4.92it/s] 75%|███████▌ | 9/12 [00:01<00:00, 4.94it/s] 83%|████████▎ | 10/12 [00:02<00:00, 4.95it/s] 92%|█████████▏| 11/12 [00:02<00:00, 4.95it/s] 100%|██████████| 12/12 [00:02<00:00, 4.95it/s] 100%|██████████| 12/12 [00:02<00:00, 4.98it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDy9vpnhfed9rme0cp9bathjd5hmStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.37
- extra_lora_scale
- 1
- num_inference_steps
- 39
{ "image": "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\n\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.37, "extra_lora_scale": 1, "num_inference_steps": 39 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\n\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.37, extra_lora_scale: 1, num_inference_steps: 39 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\n\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.37, "extra_lora_scale": 1, "num_inference_steps": 39 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\n\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.37, "extra_lora_scale": 1, "num_inference_steps": 39 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\n\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.37' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=39'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\n\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.37, "extra_lora_scale": 1, "num_inference_steps": 39 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T21:37:41.268416Z", "created_at": "2025-04-18T21:37:36.362000Z", "data_removed": false, "error": null, "id": "y9vpnhfed9rme0cp9bathjd5hm", "input": { "image": "https://replicate.delivery/pbxt/MrPSaCio00H9AUIJ0XsjpGNGMRzI29asAmYZuJVgmHKh05Ao/blackhat_variation_III.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\n\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.37, "extra_lora_scale": 1, "num_inference_steps": 39 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 63149\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:02, 5.53it/s]\n 13%|█▎ | 2/15 [00:00<00:02, 5.19it/s]\n 20%|██ | 3/15 [00:00<00:02, 5.08it/s]\n 27%|██▋ | 4/15 [00:00<00:02, 5.04it/s]\n 33%|███▎ | 5/15 [00:00<00:01, 5.00it/s]\n 40%|████ | 6/15 [00:01<00:01, 4.98it/s]\n 47%|████▋ | 7/15 [00:01<00:01, 4.97it/s]\n 53%|█████▎ | 8/15 [00:01<00:01, 4.95it/s]\n 60%|██████ | 9/15 [00:01<00:01, 4.94it/s]\n 67%|██████▋ | 10/15 [00:01<00:01, 4.95it/s]\n 73%|███████▎ | 11/15 [00:02<00:00, 4.95it/s]\n 80%|████████ | 12/15 [00:02<00:00, 4.94it/s]\n 87%|████████▋ | 13/15 [00:02<00:00, 4.93it/s]\n 93%|█████████▎| 14/15 [00:02<00:00, 4.94it/s]\n100%|██████████| 15/15 [00:03<00:00, 4.92it/s]\n100%|██████████| 15/15 [00:03<00:00, 4.97it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 3.9441525889999998, "total_time": 4.906416 }, "output": [ "https://replicate.delivery/xezq/fKR6RYqdF10qVCJVlJBjXfhF76LiiZeH88NbKkfeYhZqE8gkC/out-0.webp" ], "started_at": "2025-04-18T21:37:37.324263Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-dq3tydwcigqegl2z6fceb2o7nfhs2gzwsfopppffjsgv4h6r5icq", "get": "https://api.replicate.com/v1/predictions/y9vpnhfed9rme0cp9bathjd5hm", "cancel": "https://api.replicate.com/v1/predictions/y9vpnhfed9rme0cp9bathjd5hm/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 63149 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:02, 5.53it/s] 13%|█▎ | 2/15 [00:00<00:02, 5.19it/s] 20%|██ | 3/15 [00:00<00:02, 5.08it/s] 27%|██▋ | 4/15 [00:00<00:02, 5.04it/s] 33%|███▎ | 5/15 [00:00<00:01, 5.00it/s] 40%|████ | 6/15 [00:01<00:01, 4.98it/s] 47%|████▋ | 7/15 [00:01<00:01, 4.97it/s] 53%|█████▎ | 8/15 [00:01<00:01, 4.95it/s] 60%|██████ | 9/15 [00:01<00:01, 4.94it/s] 67%|██████▋ | 10/15 [00:01<00:01, 4.95it/s] 73%|███████▎ | 11/15 [00:02<00:00, 4.95it/s] 80%|████████ | 12/15 [00:02<00:00, 4.94it/s] 87%|████████▋ | 13/15 [00:02<00:00, 4.93it/s] 93%|█████████▎| 14/15 [00:02<00:00, 4.94it/s] 100%|██████████| 15/15 [00:03<00:00, 4.92it/s] 100%|██████████| 15/15 [00:03<00:00, 4.97it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID5yypg07eh9rm80cp9bm9h1dcpmStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard.
- go_fast
- lora_scale
- 1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.33
- extra_lora_scale
- 1
- num_inference_steps
- 43
{ "mask": "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.33, "extra_lora_scale": 1, "num_inference_steps": 43 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", image: "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", go_fast: false, lora_scale: 1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.33, extra_lora_scale: 1, num_inference_steps: 43 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": False, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.33, "extra_lora_scale": 1, "num_inference_steps": 43 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.33, "extra_lora_scale": 1, "num_inference_steps": 43 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png"' \ -i 'image="https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard."' \ -i 'go_fast=false' \ -i 'lora_scale=1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.33' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=43'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\\nBackground: professional fashion studio (charcoal gray or soft gradient).\\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model\'s physique.\\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.33, "extra_lora_scale": 1, "num_inference_steps": 43 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T21:58:25.631791Z", "created_at": "2025-04-18T21:58:21.578000Z", "data_removed": false, "error": null, "id": "5yypg07eh9rm80cp9bm9h1dcpm", "input": { "mask": "https://replicate.delivery/pbxt/MrPmFKCL2XYkc7bdoJvuL9XrwJtE0CFBnKthZz7xdcaURPen/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.33, "extra_lora_scale": 1, "num_inference_steps": 43 }, "logs": "Loaded LoRAs in 0.55s\nUsing seed: 27188\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nUse cinematic studio lighting, soft defined shadows, and a crisp depth-of-field.\nBackground: professional fashion studio (charcoal gray or soft gradient).\nPrioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique.\nPose: confident, slightly dynamic, fashion-editorial standard.\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] inpaint mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:03, 4.18it/s]\n 13%|█▎ | 2/15 [00:00<00:02, 4.61it/s]\n 20%|██ | 3/15 [00:00<00:02, 4.76it/s]\n 27%|██▋ | 4/15 [00:00<00:02, 4.84it/s]\n 33%|███▎ | 5/15 [00:01<00:02, 4.88it/s]\n 40%|████ | 6/15 [00:01<00:01, 4.91it/s]\n 47%|████▋ | 7/15 [00:01<00:01, 4.92it/s]\n 53%|█████▎ | 8/15 [00:01<00:01, 4.94it/s]\n 60%|██████ | 9/15 [00:01<00:01, 4.94it/s]\n 67%|██████▋ | 10/15 [00:02<00:01, 4.95it/s]\n 73%|███████▎ | 11/15 [00:02<00:00, 4.95it/s]\n 80%|████████ | 12/15 [00:02<00:00, 4.96it/s]\n 87%|████████▋ | 13/15 [00:02<00:00, 4.96it/s]\n 93%|█████████▎| 14/15 [00:02<00:00, 4.96it/s]\n100%|██████████| 15/15 [00:03<00:00, 5.27it/s]\n100%|██████████| 15/15 [00:03<00:00, 4.96it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 4.024188731, "total_time": 4.053791 }, "output": [ "https://replicate.delivery/xezq/hao7nkpXUezHRyfbX6vze9Y8IFIzL2BvTi4IKfpouYvHQfgkC/out-0.webp" ], "started_at": "2025-04-18T21:58:21.607602Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-onixccm56y2pnrhoeasfk6iveae6os3foxm5rqto3ymcuhucgjhq", "get": "https://api.replicate.com/v1/predictions/5yypg07eh9rm80cp9bm9h1dcpm", "cancel": "https://api.replicate.com/v1/predictions/5yypg07eh9rm80cp9bm9h1dcpm/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.55s Using seed: 27188 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image: a black off-shoulder sweater adorned with colorful geometric shapes (red, yellow, green, blue), including squares, circles, triangles, and rectangles. The sweater features long black fringe tassels hanging from the elbows. The outfit is paired with high-waisted, slim, ribbed black pants. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Use cinematic studio lighting, soft defined shadows, and a crisp depth-of-field. Background: professional fashion studio (charcoal gray or soft gradient). Prioritize ultra-realistic textile rendering, visible stitching where appropriate, and organic garment draping that matches the model's physique. Pose: confident, slightly dynamic, fashion-editorial standard. Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] inpaint mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:03, 4.18it/s] 13%|█▎ | 2/15 [00:00<00:02, 4.61it/s] 20%|██ | 3/15 [00:00<00:02, 4.76it/s] 27%|██▋ | 4/15 [00:00<00:02, 4.84it/s] 33%|███▎ | 5/15 [00:01<00:02, 4.88it/s] 40%|████ | 6/15 [00:01<00:01, 4.91it/s] 47%|████▋ | 7/15 [00:01<00:01, 4.92it/s] 53%|█████▎ | 8/15 [00:01<00:01, 4.94it/s] 60%|██████ | 9/15 [00:01<00:01, 4.94it/s] 67%|██████▋ | 10/15 [00:02<00:01, 4.95it/s] 73%|███████▎ | 11/15 [00:02<00:00, 4.95it/s] 80%|████████ | 12/15 [00:02<00:00, 4.96it/s] 87%|████████▋ | 13/15 [00:02<00:00, 4.96it/s] 93%|█████████▎| 14/15 [00:02<00:00, 4.96it/s] 100%|██████████| 15/15 [00:03<00:00, 5.27it/s] 100%|██████████| 15/15 [00:03<00:00, 4.96it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID7vvrgnwqr5rm80cp9bnbc3j920StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.05
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.35
- extra_lora_scale
- 1
- num_inference_steps
- 41
{ "mask": "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.05, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.35, "extra_lora_scale": 1, "num_inference_steps": 41 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", image: "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.05, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.35, extra_lora_scale: 1, num_inference_steps: 41 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.05, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.35, "extra_lora_scale": 1, "num_inference_steps": 41 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.05, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.35, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png"' \ -i 'image="https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.05' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.35' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=41'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.05, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.35, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T22:00:15.823382Z", "created_at": "2025-04-18T22:00:10.433000Z", "data_removed": false, "error": null, "id": "7vvrgnwqr5rm80cp9bnbc3j920", "input": { "mask": "https://replicate.delivery/pbxt/MrPnxfRrfDe7dwRhtJ6NDHSBmmvBnsXxEKSemv31njJgmKXI/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.05, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.35, "extra_lora_scale": 1, "num_inference_steps": 41 }, "logs": "free=25799175577600\nDownloading weights\n2025-04-18T22:00:10Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpvmyz61vc/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-18T22:00:10Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-18T22:00:11Z | INFO | [ Complete ] dest=/tmp/tmpvmyz61vc/weights size=\"172 MB\" total_elapsed=0.581s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 0.60s\nLoaded LoRAs in 1.14s\nUsing seed: 34759\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] inpaint mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:03, 4.12it/s]\n 13%|█▎ | 2/15 [00:00<00:02, 4.47it/s]\n 20%|██ | 3/15 [00:00<00:02, 4.61it/s]\n 27%|██▋ | 4/15 [00:00<00:02, 4.67it/s]\n 33%|███▎ | 5/15 [00:01<00:02, 4.71it/s]\n 40%|████ | 6/15 [00:01<00:01, 4.73it/s]\n 47%|████▋ | 7/15 [00:01<00:01, 4.74it/s]\n 53%|█████▎ | 8/15 [00:01<00:01, 4.75it/s]\n 60%|██████ | 9/15 [00:01<00:01, 4.76it/s]\n 67%|██████▋ | 10/15 [00:02<00:01, 4.77it/s]\n 73%|███████▎ | 11/15 [00:02<00:00, 4.77it/s]\n 80%|████████ | 12/15 [00:02<00:00, 4.77it/s]\n 87%|████████▋ | 13/15 [00:02<00:00, 4.77it/s]\n 93%|█████████▎| 14/15 [00:02<00:00, 4.78it/s]\n100%|██████████| 15/15 [00:03<00:00, 5.02it/s]\n100%|██████████| 15/15 [00:03<00:00, 4.78it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 5.30822834, "total_time": 5.390382 }, "output": [ "https://replicate.delivery/xezq/XmegblOjra1AJaQ5c1ac6gWlvOthTIhehn1Kxchhrf4fWfgkC/out-0.webp" ], "started_at": "2025-04-18T22:00:10.515154Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-bcbcowu2thankpfljxixcsmxugfknvrcp4ab4nyg7kqasxdpqw5q", "get": "https://api.replicate.com/v1/predictions/7vvrgnwqr5rm80cp9bnbc3j920", "cancel": "https://api.replicate.com/v1/predictions/7vvrgnwqr5rm80cp9bnbc3j920/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25799175577600 Downloading weights 2025-04-18T22:00:10Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpvmyz61vc/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-18T22:00:10Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-18T22:00:11Z | INFO | [ Complete ] dest=/tmp/tmpvmyz61vc/weights size="172 MB" total_elapsed=0.581s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 0.60s Loaded LoRAs in 1.14s Using seed: 34759 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] inpaint mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:03, 4.12it/s] 13%|█▎ | 2/15 [00:00<00:02, 4.47it/s] 20%|██ | 3/15 [00:00<00:02, 4.61it/s] 27%|██▋ | 4/15 [00:00<00:02, 4.67it/s] 33%|███▎ | 5/15 [00:01<00:02, 4.71it/s] 40%|████ | 6/15 [00:01<00:01, 4.73it/s] 47%|████▋ | 7/15 [00:01<00:01, 4.74it/s] 53%|█████▎ | 8/15 [00:01<00:01, 4.75it/s] 60%|██████ | 9/15 [00:01<00:01, 4.76it/s] 67%|██████▋ | 10/15 [00:02<00:01, 4.77it/s] 73%|███████▎ | 11/15 [00:02<00:00, 4.77it/s] 80%|████████ | 12/15 [00:02<00:00, 4.77it/s] 87%|████████▋ | 13/15 [00:02<00:00, 4.77it/s] 93%|█████████▎| 14/15 [00:02<00:00, 4.78it/s] 100%|██████████| 15/15 [00:03<00:00, 5.02it/s] 100%|██████████| 15/15 [00:03<00:00, 4.78it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID26t6y02xgxrme0cp9bv885twsmStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.25
- extra_lora_scale
- 1
- num_inference_steps
- 41
{ "mask": "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.25, "extra_lora_scale": 1, "num_inference_steps": 41 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", image: "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.25, extra_lora_scale: 1, num_inference_steps: 41 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.25, "extra_lora_scale": 1, "num_inference_steps": 41 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.25, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png"' \ -i 'image="https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.25' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=41'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.25, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T22:13:09.067307Z", "created_at": "2025-04-18T22:13:01.959000Z", "data_removed": false, "error": null, "id": "26t6y02xgxrme0cp9bv885twsm", "input": { "mask": "https://replicate.delivery/pbxt/MrQ093GT66DBdXqvFi5A8anLNgjdoPY0Fw2J73lX4ufFRtKu/Group%2037.png", "image": "https://replicate.delivery/pbxt/MrPYIsb9YFBHFsb647jGdUuIsKAW7IPSZP8HVhnMlOQlOy9G/Black%20Hat%20_%20v2.jpeg", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.25, "extra_lora_scale": 1, "num_inference_steps": 41 }, "logs": "free=25293505998848\nDownloading weights\n2025-04-18T22:13:02Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpmwpy1vq6/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-18T22:13:02Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-18T22:13:05Z | INFO | [ Complete ] dest=/tmp/tmpmwpy1vq6/weights size=\"172 MB\" total_elapsed=3.162s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 3.18s\nLoaded LoRAs in 3.73s\nUsing seed: 64606\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 720x1080\n[!] Resizing input image from 720x1080 to 720x1088\n[!] inpaint mode\n 0%| | 0/11 [00:00<?, ?it/s]\n 9%|▉ | 1/11 [00:00<00:02, 4.11it/s]\n 18%|█▊ | 2/11 [00:00<00:02, 4.46it/s]\n 27%|██▋ | 3/11 [00:00<00:01, 4.59it/s]\n 36%|███▋ | 4/11 [00:00<00:01, 4.66it/s]\n 45%|████▌ | 5/11 [00:01<00:01, 4.70it/s]\n 55%|█████▍ | 6/11 [00:01<00:01, 4.71it/s]\n 64%|██████▎ | 7/11 [00:01<00:00, 4.73it/s]\n 73%|███████▎ | 8/11 [00:01<00:00, 4.74it/s]\n 82%|████████▏ | 9/11 [00:01<00:00, 4.75it/s]\n 91%|█████████ | 10/11 [00:02<00:00, 4.76it/s]\n100%|██████████| 11/11 [00:02<00:00, 5.01it/s]\n100%|██████████| 11/11 [00:02<00:00, 4.76it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 7.062979686, "total_time": 7.108307 }, "output": [ "https://replicate.delivery/xezq/SLNymrjgwKpJB9CliacJ5lKDJ53elcfBRlAiGjWLbpc1BIkUA/out-0.webp" ], "started_at": "2025-04-18T22:13:02.004327Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-h36hqlxr5bitlgsayab66vf6vtwkzhqfvnfixffyc2rlyjtgoz5a", "get": "https://api.replicate.com/v1/predictions/26t6y02xgxrme0cp9bv885twsm", "cancel": "https://api.replicate.com/v1/predictions/26t6y02xgxrme0cp9bv885twsm/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25293505998848 Downloading weights 2025-04-18T22:13:02Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpmwpy1vq6/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-18T22:13:02Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-18T22:13:05Z | INFO | [ Complete ] dest=/tmp/tmpmwpy1vq6/weights size="172 MB" total_elapsed=3.162s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 3.18s Loaded LoRAs in 3.73s Using seed: 64606 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 720x1080 [!] Resizing input image from 720x1080 to 720x1088 [!] inpaint mode 0%| | 0/11 [00:00<?, ?it/s] 9%|▉ | 1/11 [00:00<00:02, 4.11it/s] 18%|█▊ | 2/11 [00:00<00:02, 4.46it/s] 27%|██▋ | 3/11 [00:00<00:01, 4.59it/s] 36%|███▋ | 4/11 [00:00<00:01, 4.66it/s] 45%|████▌ | 5/11 [00:01<00:01, 4.70it/s] 55%|█████▍ | 6/11 [00:01<00:01, 4.71it/s] 64%|██████▎ | 7/11 [00:01<00:00, 4.73it/s] 73%|███████▎ | 8/11 [00:01<00:00, 4.74it/s] 82%|████████▏ | 9/11 [00:01<00:00, 4.75it/s] 91%|█████████ | 10/11 [00:02<00:00, 4.76it/s] 100%|██████████| 11/11 [00:02<00:00, 5.01it/s] 100%|██████████| 11/11 [00:02<00:00, 4.76it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDxg58sa0wchrme0cp9c0rmw8cfmStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.1
- output_quality
- 80
- prompt_strength
- 0.27
- extra_lora_scale
- 1
- num_inference_steps
- 41
{ "mask": "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 0.27, "extra_lora_scale": 1, "num_inference_steps": 41 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", image: "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.1, output_quality: 80, prompt_strength: 0.27, extra_lora_scale: 1, num_inference_steps: 41 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 0.27, "extra_lora_scale": 1, "num_inference_steps": 41 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 0.27, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png"' \ -i 'image="https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.1' \ -i 'output_quality=80' \ -i 'prompt_strength=0.27' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=41'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 0.27, "extra_lora_scale": 1, "num_inference_steps": 41 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T22:24:55.031425Z", "created_at": "2025-04-18T22:24:46.180000Z", "data_removed": false, "error": null, "id": "xg58sa0wchrme0cp9c0rmw8cfm", "input": { "mask": "https://replicate.delivery/pbxt/MrQBAW0HjNuDRIrqTq4x9DY1VBU96eSi95da0BfAR6pWGRQW/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQBA20i6soeK469T2UEsSvBicdrLg3qOGvb3ObnetMSco0x/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 0.27, "extra_lora_scale": 1, "num_inference_steps": 41 }, "logs": "free=25461879599104\nDownloading weights\n2025-04-18T22:24:46Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp7l8d1a9b/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-18T22:24:46Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-18T22:24:48Z | INFO | [ Complete ] dest=/tmp/tmp7l8d1a9b/weights size=\"172 MB\" total_elapsed=1.102s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 1.12s\nLoaded LoRAs in 1.69s\nUsing seed: 62457\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] inpaint mode\n 0%| | 0/12 [00:00<?, ?it/s]\n 8%|▊ | 1/12 [00:00<00:04, 2.38it/s]\n 17%|█▋ | 2/12 [00:00<00:03, 2.63it/s]\n 25%|██▌ | 3/12 [00:01<00:03, 2.72it/s]\n 33%|███▎ | 4/12 [00:01<00:02, 2.76it/s]\n 42%|████▏ | 5/12 [00:01<00:02, 2.78it/s]\n 50%|█████ | 6/12 [00:02<00:02, 2.80it/s]\n 58%|█████▊ | 7/12 [00:02<00:01, 2.80it/s]\n 67%|██████▋ | 8/12 [00:02<00:01, 2.81it/s]\n 75%|███████▌ | 9/12 [00:03<00:01, 2.82it/s]\n 83%|████████▎ | 10/12 [00:03<00:00, 2.82it/s]\n 92%|█████████▏| 11/12 [00:03<00:00, 2.82it/s]\n100%|██████████| 12/12 [00:04<00:00, 2.99it/s]\n100%|██████████| 12/12 [00:04<00:00, 2.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.414293168, "total_time": 8.851425 }, "output": [ "https://replicate.delivery/xezq/ScnIsuOLlp4uJZCyDgcktaLRlfCnhUbR9i6Wz615CvybGESKA/out-0.webp" ], "started_at": "2025-04-18T22:24:46.617132Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-av7z5wvtwjsuqocw3etwo5tymipzbegesjqdf25k2dmnsfpbbfva", "get": "https://api.replicate.com/v1/predictions/xg58sa0wchrme0cp9c0rmw8cfm", "cancel": "https://api.replicate.com/v1/predictions/xg58sa0wchrme0cp9c0rmw8cfm/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25461879599104 Downloading weights 2025-04-18T22:24:46Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp7l8d1a9b/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-18T22:24:46Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-18T22:24:48Z | INFO | [ Complete ] dest=/tmp/tmp7l8d1a9b/weights size="172 MB" total_elapsed=1.102s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 1.12s Loaded LoRAs in 1.69s Using seed: 62457 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] inpaint mode 0%| | 0/12 [00:00<?, ?it/s] 8%|▊ | 1/12 [00:00<00:04, 2.38it/s] 17%|█▋ | 2/12 [00:00<00:03, 2.63it/s] 25%|██▌ | 3/12 [00:01<00:03, 2.72it/s] 33%|███▎ | 4/12 [00:01<00:02, 2.76it/s] 42%|████▏ | 5/12 [00:01<00:02, 2.78it/s] 50%|█████ | 6/12 [00:02<00:02, 2.80it/s] 58%|█████▊ | 7/12 [00:02<00:01, 2.80it/s] 67%|██████▋ | 8/12 [00:02<00:01, 2.81it/s] 75%|███████▌ | 9/12 [00:03<00:01, 2.82it/s] 83%|████████▎ | 10/12 [00:03<00:00, 2.82it/s] 92%|█████████▏| 11/12 [00:03<00:00, 2.82it/s] 100%|██████████| 12/12 [00:04<00:00, 2.99it/s] 100%|██████████| 12/12 [00:04<00:00, 2.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDy3zhnzbdcdrm80cp9c5tgmmkhgStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.02
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.3
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "mask": "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", image: "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.02, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.3, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png"' \ -i 'image="https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.02' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.3' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T22:36:10.321996Z", "created_at": "2025-04-18T22:36:02.275000Z", "data_removed": false, "error": null, "id": "y3zhnzbdcdrm80cp9c5tgmmkhg", "input": { "mask": "https://replicate.delivery/pbxt/MrQLsIxoUJH0tA8P5AL1GAaDIAhQ67PhtNB8Q7MMwfjnE3HA/Group%2037-2.png", "image": "https://replicate.delivery/pbxt/MrQLsCuIqfGAy4ireFYpDt8wwhXzrXoqL9NhH8DvO6TmqAj6/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.3, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 2.52s\nUsing seed: 5005\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] inpaint mode\n 0%| | 0/12 [00:00<?, ?it/s]\n 8%|▊ | 1/12 [00:00<00:04, 2.39it/s]\n 17%|█▋ | 2/12 [00:00<00:03, 2.64it/s]\n 25%|██▌ | 3/12 [00:01<00:03, 2.72it/s]\n 33%|███▎ | 4/12 [00:01<00:02, 2.76it/s]\n 42%|████▏ | 5/12 [00:01<00:02, 2.79it/s]\n 50%|█████ | 6/12 [00:02<00:02, 2.81it/s]\n 58%|█████▊ | 7/12 [00:02<00:01, 2.81it/s]\n 67%|██████▋ | 8/12 [00:02<00:01, 2.82it/s]\n 75%|███████▌ | 9/12 [00:03<00:01, 2.82it/s]\n 83%|████████▎ | 10/12 [00:03<00:00, 2.82it/s]\n 92%|█████████▏| 11/12 [00:03<00:00, 2.81it/s]\n100%|██████████| 12/12 [00:04<00:00, 2.99it/s]\n100%|██████████| 12/12 [00:04<00:00, 2.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 7.519323651, "total_time": 8.046996 }, "output": [ "https://replicate.delivery/xezq/cGLQ1HRkOmbgLlLNNs47A4BEEGZC5ZeeQmTvfO97mhX0uQIpA/out-0.webp" ], "started_at": "2025-04-18T22:36:02.802673Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-sn6xyonq4mkks3iqsuwltcmki47a7lh7a2slkl54bnaiqxxd2tpq", "get": "https://api.replicate.com/v1/predictions/y3zhnzbdcdrm80cp9c5tgmmkhg", "cancel": "https://api.replicate.com/v1/predictions/y3zhnzbdcdrm80cp9c5tgmmkhg/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 2.52s Using seed: 5005 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] inpaint mode 0%| | 0/12 [00:00<?, ?it/s] 8%|▊ | 1/12 [00:00<00:04, 2.39it/s] 17%|█▋ | 2/12 [00:00<00:03, 2.64it/s] 25%|██▌ | 3/12 [00:01<00:03, 2.72it/s] 33%|███▎ | 4/12 [00:01<00:02, 2.76it/s] 42%|████▏ | 5/12 [00:01<00:02, 2.79it/s] 50%|█████ | 6/12 [00:02<00:02, 2.81it/s] 58%|█████▊ | 7/12 [00:02<00:01, 2.81it/s] 67%|██████▋ | 8/12 [00:02<00:01, 2.82it/s] 75%|███████▌ | 9/12 [00:03<00:01, 2.82it/s] 83%|████████▎ | 10/12 [00:03<00:00, 2.82it/s] 92%|█████████▏| 11/12 [00:03<00:00, 2.81it/s] 100%|██████████| 12/12 [00:04<00:00, 2.99it/s] 100%|██████████| 12/12 [00:04<00:00, 2.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID7k8bcy1mwsrmc0cp9c6vfjzbr8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- seed
- 5005
- model
- dev
- prompt
- Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.
- go_fast
- lora_scale
- 1.02
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2
- output_quality
- 80
- prompt_strength
- 0.29
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "mask": "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", "seed": 5005, "image": "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.29, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { mask: "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", seed: 5005, image: "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", model: "dev", prompt: "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", go_fast: false, lora_scale: 1.02, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2, output_quality: 80, prompt_strength: 0.29, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "mask": "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", "seed": 5005, "image": "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": False, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.29, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "mask": "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", "seed": 5005, "image": "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.29, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'mask="https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png"' \ -i 'seed=5005' \ -i 'image="https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior."' \ -i 'go_fast=false' \ -i 'lora_scale=1.02' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2' \ -i 'output_quality=80' \ -i 'prompt_strength=0.29' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "mask": "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", "seed": 5005, "image": "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\\nMaintain TOKMAHI\'s facial identity, natural skin texture, and hairstyle.\\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.29, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-18T22:38:14.353705Z", "created_at": "2025-04-18T22:37:58.886000Z", "data_removed": false, "error": null, "id": "7k8bcy1mwsrmc0cp9c6vfjzbr8", "input": { "mask": "https://replicate.delivery/pbxt/MrQNLKeynyGLufoEBWgk4FCUW0chVeqxbABBMzJmXrgynX09/Group%2037-2.png", "seed": 5005, "image": "https://replicate.delivery/pbxt/MrQNLMtUZB0MrA5W6RKncTXjkEDOTABykxrSsAbphgJmaXCk/ChatGPT%20Image%20Apr%2018%2C%202025%20at%2010_24_18%20PM.png", "model": "dev", "prompt": "Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2, "output_quality": 80, "prompt_strength": 0.29, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "free=25501186691072\nDownloading weights\n2025-04-18T22:37:59Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmptccl8d88/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-18T22:37:59Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-18T22:38:08Z | INFO | [ Complete ] dest=/tmp/tmptccl8d88/weights size=\"172 MB\" total_elapsed=8.849s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 8.87s\nLoaded LoRAs in 9.42s\nUsing seed: 5005\nPrompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat.\nMaintain TOKMAHI's facial identity, natural skin texture, and hairstyle.\nStrictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations.\nAlso replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image.\nThe garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior.\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] inpaint mode\n 0%| | 0/12 [00:00<?, ?it/s]\n 8%|▊ | 1/12 [00:00<00:04, 2.24it/s]\n 17%|█▋ | 2/12 [00:00<00:03, 2.56it/s]\n 25%|██▌ | 3/12 [00:01<00:03, 2.69it/s]\n 33%|███▎ | 4/12 [00:01<00:02, 2.75it/s]\n 42%|████▏ | 5/12 [00:01<00:02, 2.79it/s]\n 50%|█████ | 6/12 [00:02<00:02, 2.81it/s]\n 58%|█████▊ | 7/12 [00:02<00:01, 2.82it/s]\n 67%|██████▋ | 8/12 [00:02<00:01, 2.83it/s]\n 75%|███████▌ | 9/12 [00:03<00:01, 2.83it/s]\n 83%|████████▎ | 10/12 [00:03<00:00, 2.83it/s]\n 92%|█████████▏| 11/12 [00:03<00:00, 2.84it/s]\n100%|██████████| 12/12 [00:04<00:00, 3.00it/s]\n100%|██████████| 12/12 [00:04<00:00, 2.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 15.459291768, "total_time": 15.467705 }, "output": [ "https://replicate.delivery/xezq/TRBDBeScMs0cIq1k2IwvJfwDu3Sh9HOsApERD58l2i1WZIkUA/out-0.webp" ], "started_at": "2025-04-18T22:37:58.894413Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-kp5juvt452qfao4z2clxqse7g7qpubtk6nkhxlv5t56o6olsmpwa", "get": "https://api.replicate.com/v1/predictions/7k8bcy1mwsrmc0cp9c6vfjzbr8", "cancel": "https://api.replicate.com/v1/predictions/7k8bcy1mwsrmc0cp9c6vfjzbr8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25501186691072 Downloading weights 2025-04-18T22:37:59Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmptccl8d88/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-18T22:37:59Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-18T22:38:08Z | INFO | [ Complete ] dest=/tmp/tmptccl8d88/weights size="172 MB" total_elapsed=8.849s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 8.87s Loaded LoRAs in 9.42s Using seed: 5005 Prompt: Generate an ultra-detailed full-body portrait of TOKMAHI, styled in the outfit shown in the uploaded reference image. TOKMAHI must wear a wide-brimmed, flat-crowned black hat. Maintain TOKMAHI's facial identity, natural skin texture, and hairstyle. Strictly replicate the same fabric material, clothing style, silhouette, patterns, embroidery, and construction details from the reference without alterations. Also replicate the body type, height, posture, and overall proportions of the model shown wearing the outfit in the reference image. The garment must fit TOKMAHI exactly as it fits the reference model, preserving proportions, silhouette, and garment behavior. Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] inpaint mode 0%| | 0/12 [00:00<?, ?it/s] 8%|▊ | 1/12 [00:00<00:04, 2.24it/s] 17%|█▋ | 2/12 [00:00<00:03, 2.56it/s] 25%|██▌ | 3/12 [00:01<00:03, 2.69it/s] 33%|███▎ | 4/12 [00:01<00:02, 2.75it/s] 42%|████▏ | 5/12 [00:01<00:02, 2.79it/s] 50%|█████ | 6/12 [00:02<00:02, 2.81it/s] 58%|█████▊ | 7/12 [00:02<00:01, 2.82it/s] 67%|██████▋ | 8/12 [00:02<00:01, 2.83it/s] 75%|███████▌ | 9/12 [00:03<00:01, 2.83it/s] 83%|████████▎ | 10/12 [00:03<00:00, 2.83it/s] 92%|█████████▏| 11/12 [00:03<00:00, 2.84it/s] 100%|██████████| 12/12 [00:04<00:00, 3.00it/s] 100%|██████████| 12/12 [00:04<00:00, 2.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDbmyw8633zhrm80cp9shrzgs544StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer Style: ethereal, dreamlike Composition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges Lighting: soft golden shafts illuminating her features, delicate rim highlights on hair Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 0.88
- extra_lora_scale
- 1
- num_inference_steps
- 34
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.88, "extra_lora_scale": 1, "num_inference_steps": 34 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 0.88, extra_lora_scale: 1, num_inference_steps: 34 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.88, "extra_lora_scale": 1, "num_inference_steps": 34 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.88, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=0.88' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=34'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.88, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T14:11:08.410859Z", "created_at": "2025-04-19T14:10:58.684000Z", "data_removed": false, "error": null, "id": "bmyw8633zhrm80cp9shrzgs544", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 0.88, "extra_lora_scale": 1, "num_inference_steps": 34 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 36155\nPrompt:\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer\nStyle: ethereal, dreamlike\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/34 [00:00<?, ?it/s]\n 3%|▎ | 1/34 [00:00<00:08, 3.81it/s]\n 6%|▌ | 2/34 [00:00<00:07, 4.32it/s]\n 9%|▉ | 3/34 [00:00<00:07, 4.07it/s]\n 12%|█▏ | 4/34 [00:00<00:07, 3.97it/s]\n 15%|█▍ | 5/34 [00:01<00:07, 3.92it/s]\n 18%|█▊ | 6/34 [00:01<00:07, 3.88it/s]\n 21%|██ | 7/34 [00:01<00:07, 3.86it/s]\n 24%|██▎ | 8/34 [00:02<00:06, 3.84it/s]\n 26%|██▋ | 9/34 [00:02<00:06, 3.84it/s]\n 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s]\n 32%|███▏ | 11/34 [00:02<00:06, 3.83it/s]\n 35%|███▌ | 12/34 [00:03<00:05, 3.83it/s]\n 38%|███▊ | 13/34 [00:03<00:05, 3.82it/s]\n 41%|████ | 14/34 [00:03<00:05, 3.82it/s]\n 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 16/34 [00:04<00:04, 3.81it/s]\n 50%|█████ | 17/34 [00:04<00:04, 3.80it/s]\n 53%|█████▎ | 18/34 [00:04<00:04, 3.81it/s]\n 56%|█████▌ | 19/34 [00:04<00:03, 3.81it/s]\n 59%|█████▉ | 20/34 [00:05<00:03, 3.81it/s]\n 62%|██████▏ | 21/34 [00:05<00:03, 3.81it/s]\n 65%|██████▍ | 22/34 [00:05<00:03, 3.81it/s]\n 68%|██████▊ | 23/34 [00:05<00:02, 3.81it/s]\n 71%|███████ | 24/34 [00:06<00:02, 3.81it/s]\n 74%|███████▎ | 25/34 [00:06<00:02, 3.81it/s]\n 76%|███████▋ | 26/34 [00:06<00:02, 3.80it/s]\n 79%|███████▉ | 27/34 [00:07<00:01, 3.81it/s]\n 82%|████████▏ | 28/34 [00:07<00:01, 3.81it/s]\n 85%|████████▌ | 29/34 [00:07<00:01, 3.81it/s]\n 88%|████████▊ | 30/34 [00:07<00:01, 3.81it/s]\n 91%|█████████ | 31/34 [00:08<00:00, 3.81it/s]\n 94%|█████████▍| 32/34 [00:08<00:00, 3.81it/s]\n 97%|█████████▋| 33/34 [00:08<00:00, 3.81it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.81it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.717953693, "total_time": 9.726859 }, "output": [ "https://replicate.delivery/xezq/Qe817hfOC6jzfJPXAlqHxEfxMWEukFxP2XRLbyQOeLIifgFJF/out-0.webp" ], "started_at": "2025-04-19T14:10:58.692905Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-43h5d6lmadwm2ys6qzs6m6v6c6pii7kbgy2xbk7mrqe3r5eacjfa", "get": "https://api.replicate.com/v1/predictions/bmyw8633zhrm80cp9shrzgs544", "cancel": "https://api.replicate.com/v1/predictions/bmyw8633zhrm80cp9shrzgs544/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 36155 Prompt: Subject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer Style: ethereal, dreamlike Composition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges Lighting: soft golden shafts illuminating her features, delicate rim highlights on hair Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/34 [00:00<?, ?it/s] 3%|▎ | 1/34 [00:00<00:08, 3.81it/s] 6%|▌ | 2/34 [00:00<00:07, 4.32it/s] 9%|▉ | 3/34 [00:00<00:07, 4.07it/s] 12%|█▏ | 4/34 [00:00<00:07, 3.97it/s] 15%|█▍ | 5/34 [00:01<00:07, 3.92it/s] 18%|█▊ | 6/34 [00:01<00:07, 3.88it/s] 21%|██ | 7/34 [00:01<00:07, 3.86it/s] 24%|██▎ | 8/34 [00:02<00:06, 3.84it/s] 26%|██▋ | 9/34 [00:02<00:06, 3.84it/s] 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s] 32%|███▏ | 11/34 [00:02<00:06, 3.83it/s] 35%|███▌ | 12/34 [00:03<00:05, 3.83it/s] 38%|███▊ | 13/34 [00:03<00:05, 3.82it/s] 41%|████ | 14/34 [00:03<00:05, 3.82it/s] 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s] 47%|████▋ | 16/34 [00:04<00:04, 3.81it/s] 50%|█████ | 17/34 [00:04<00:04, 3.80it/s] 53%|█████▎ | 18/34 [00:04<00:04, 3.81it/s] 56%|█████▌ | 19/34 [00:04<00:03, 3.81it/s] 59%|█████▉ | 20/34 [00:05<00:03, 3.81it/s] 62%|██████▏ | 21/34 [00:05<00:03, 3.81it/s] 65%|██████▍ | 22/34 [00:05<00:03, 3.81it/s] 68%|██████▊ | 23/34 [00:05<00:02, 3.81it/s] 71%|███████ | 24/34 [00:06<00:02, 3.81it/s] 74%|███████▎ | 25/34 [00:06<00:02, 3.81it/s] 76%|███████▋ | 26/34 [00:06<00:02, 3.80it/s] 79%|███████▉ | 27/34 [00:07<00:01, 3.81it/s] 82%|████████▏ | 28/34 [00:07<00:01, 3.81it/s] 85%|████████▌ | 29/34 [00:07<00:01, 3.81it/s] 88%|████████▊ | 30/34 [00:07<00:01, 3.81it/s] 91%|█████████ | 31/34 [00:08<00:00, 3.81it/s] 94%|█████████▍| 32/34 [00:08<00:00, 3.81it/s] 97%|█████████▋| 33/34 [00:08<00:00, 3.81it/s] 100%|██████████| 34/34 [00:08<00:00, 3.81it/s] 100%|██████████| 34/34 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDn47ck5c93hrm80cp9sm9fpf2wrStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer Style: ethereal, dreamlike Composition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges Lighting: soft golden shafts illuminating her features, delicate rim highlights on hair Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.1
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 34
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.1, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 34 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.1' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=34'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \\nStyle: ethereal, dreamlike \\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T14:16:45.754527Z", "created_at": "2025-04-19T14:16:35.868000Z", "data_removed": false, "error": null, "id": "n47ck5c93hrm80cp9sm9fpf2wr", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer \nStyle: ethereal, dreamlike \nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges \nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.1, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 40945\nPrompt:\nSubject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer\nStyle: ethereal, dreamlike\nComposition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges\nLighting: soft golden shafts illuminating her features, delicate rim highlights on hair\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate\nTechnical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/34 [00:00<?, ?it/s]\n 3%|▎ | 1/34 [00:00<00:08, 3.82it/s]\n 6%|▌ | 2/34 [00:00<00:07, 4.33it/s]\n 9%|▉ | 3/34 [00:00<00:07, 4.09it/s]\n 12%|█▏ | 4/34 [00:00<00:07, 3.97it/s]\n 15%|█▍ | 5/34 [00:01<00:07, 3.91it/s]\n 18%|█▊ | 6/34 [00:01<00:07, 3.88it/s]\n 21%|██ | 7/34 [00:01<00:06, 3.86it/s]\n 24%|██▎ | 8/34 [00:02<00:06, 3.85it/s]\n 26%|██▋ | 9/34 [00:02<00:06, 3.84it/s]\n 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s]\n 32%|███▏ | 11/34 [00:02<00:06, 3.83it/s]\n 35%|███▌ | 12/34 [00:03<00:05, 3.83it/s]\n 38%|███▊ | 13/34 [00:03<00:05, 3.83it/s]\n 41%|████ | 14/34 [00:03<00:05, 3.82it/s]\n 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 16/34 [00:04<00:04, 3.82it/s]\n 50%|█████ | 17/34 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 18/34 [00:04<00:04, 3.82it/s]\n 56%|█████▌ | 19/34 [00:04<00:03, 3.82it/s]\n 59%|█████▉ | 20/34 [00:05<00:03, 3.82it/s]\n 62%|██████▏ | 21/34 [00:05<00:03, 3.82it/s]\n 65%|██████▍ | 22/34 [00:05<00:03, 3.82it/s]\n 68%|██████▊ | 23/34 [00:05<00:02, 3.82it/s]\n 71%|███████ | 24/34 [00:06<00:02, 3.82it/s]\n 74%|███████▎ | 25/34 [00:06<00:02, 3.82it/s]\n 76%|███████▋ | 26/34 [00:06<00:02, 3.82it/s]\n 79%|███████▉ | 27/34 [00:07<00:01, 3.82it/s]\n 82%|████████▏ | 28/34 [00:07<00:01, 3.82it/s]\n 85%|████████▌ | 29/34 [00:07<00:01, 3.82it/s]\n 88%|████████▊ | 30/34 [00:07<00:01, 3.82it/s]\n 91%|█████████ | 31/34 [00:08<00:00, 3.82it/s]\n 94%|█████████▍| 32/34 [00:08<00:00, 3.82it/s]\n 97%|█████████▋| 33/34 [00:08<00:00, 3.82it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.81it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.706684522, "total_time": 9.886527 }, "output": [ "https://replicate.delivery/xezq/TGaRPn7U7Z5rP5A1VbTC8VEfen3J740TB71eX5aQfHN2kYRSB/out-0.webp" ], "started_at": "2025-04-19T14:16:36.047842Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-5b2kx3e5tjsirdtm7fxz32xd4tkmfsjtrspjsyomekn3t7ckhnoq", "get": "https://api.replicate.com/v1/predictions/n47ck5c93hrm80cp9sm9fpf2wr", "cancel": "https://api.replicate.com/v1/predictions/n47ck5c93hrm80cp9sm9fpf2wr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 40945 Prompt: Subject: TOKMAHI’s face and upper torso filling the frame, eyes locked on the viewer Style: ethereal, dreamlike Composition: extreme close‑up, tight crop, inverted gravity hinted only by floating pastel clouds at the edges Lighting: soft golden shafts illuminating her features, delicate rim highlights on hair Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on her eyes and skin texture, slight motion blur on stray cloud wisps Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/34 [00:00<?, ?it/s] 3%|▎ | 1/34 [00:00<00:08, 3.82it/s] 6%|▌ | 2/34 [00:00<00:07, 4.33it/s] 9%|▉ | 3/34 [00:00<00:07, 4.09it/s] 12%|█▏ | 4/34 [00:00<00:07, 3.97it/s] 15%|█▍ | 5/34 [00:01<00:07, 3.91it/s] 18%|█▊ | 6/34 [00:01<00:07, 3.88it/s] 21%|██ | 7/34 [00:01<00:06, 3.86it/s] 24%|██▎ | 8/34 [00:02<00:06, 3.85it/s] 26%|██▋ | 9/34 [00:02<00:06, 3.84it/s] 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s] 32%|███▏ | 11/34 [00:02<00:06, 3.83it/s] 35%|███▌ | 12/34 [00:03<00:05, 3.83it/s] 38%|███▊ | 13/34 [00:03<00:05, 3.83it/s] 41%|████ | 14/34 [00:03<00:05, 3.82it/s] 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s] 47%|████▋ | 16/34 [00:04<00:04, 3.82it/s] 50%|█████ | 17/34 [00:04<00:04, 3.82it/s] 53%|█████▎ | 18/34 [00:04<00:04, 3.82it/s] 56%|█████▌ | 19/34 [00:04<00:03, 3.82it/s] 59%|█████▉ | 20/34 [00:05<00:03, 3.82it/s] 62%|██████▏ | 21/34 [00:05<00:03, 3.82it/s] 65%|██████▍ | 22/34 [00:05<00:03, 3.82it/s] 68%|██████▊ | 23/34 [00:05<00:02, 3.82it/s] 71%|███████ | 24/34 [00:06<00:02, 3.82it/s] 74%|███████▎ | 25/34 [00:06<00:02, 3.82it/s] 76%|███████▋ | 26/34 [00:06<00:02, 3.82it/s] 79%|███████▉ | 27/34 [00:07<00:01, 3.82it/s] 82%|████████▏ | 28/34 [00:07<00:01, 3.82it/s] 85%|████████▌ | 29/34 [00:07<00:01, 3.82it/s] 88%|████████▊ | 30/34 [00:07<00:01, 3.82it/s] 91%|█████████ | 31/34 [00:08<00:00, 3.82it/s] 94%|█████████▍| 32/34 [00:08<00:00, 3.82it/s] 97%|█████████▋| 33/34 [00:08<00:00, 3.82it/s] 100%|██████████| 34/34 [00:08<00:00, 3.81it/s] 100%|██████████| 34/34 [00:08<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDgtsbdf7af5rma0cp9sp8v210ggStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", go_fast: false, lora_scale: 1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", "go_fast": False, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60"' \ -i 'go_fast=false' \ -i 'lora_scale=1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T14:21:37.993614Z", "created_at": "2025-04-19T14:21:22.937000Z", "data_removed": false, "error": null, "id": "gtsbdf7af5rma0cp9sp8v210gg", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 10933\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.84it/s]\n 6%|▋ | 2/32 [00:00<00:06, 4.35it/s]\n 9%|▉ | 3/32 [00:00<00:07, 4.10it/s]\n 12%|█▎ | 4/32 [00:00<00:06, 4.00it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.95it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.91it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.89it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.88it/s]\n 28%|██▊ | 9/32 [00:02<00:05, 3.88it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.87it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.87it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.87it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.86it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.86it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.86it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.85it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.85it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.85it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.85it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.85it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.85it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.85it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.85it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.85it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.85it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.86it/s]\n 84%|████████▍ | 27/32 [00:06<00:01, 3.86it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.86it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.85it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.85it/s]\n 97%|█████████▋| 31/32 [00:07<00:00, 3.85it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.85it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.88it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.100546841, "total_time": 15.056614 }, "output": [ "https://replicate.delivery/xezq/KRZlJ0HjR76wKZAxzNF9keraCGoDwiB4gdjWsadyHAz4GLSKA/out-0.webp" ], "started_at": "2025-04-19T14:21:28.893067Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-nnxeu3tmt7g7gn37ha3a7xfi3oyq63fqqynou4srcq4oxx23cajq", "get": "https://api.replicate.com/v1/predictions/gtsbdf7af5rma0cp9sp8v210gg", "cancel": "https://api.replicate.com/v1/predictions/gtsbdf7af5rma0cp9sp8v210gg/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 10933 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining hair edges Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.84it/s] 6%|▋ | 2/32 [00:00<00:06, 4.35it/s] 9%|▉ | 3/32 [00:00<00:07, 4.10it/s] 12%|█▎ | 4/32 [00:00<00:06, 4.00it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.95it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.91it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.89it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.88it/s] 28%|██▊ | 9/32 [00:02<00:05, 3.88it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.87it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.87it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.87it/s] 41%|████ | 13/32 [00:03<00:04, 3.86it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.86it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.86it/s] 50%|█████ | 16/32 [00:04<00:04, 3.85it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.85it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.85it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.85it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.85it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.85it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.85it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.85it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.85it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.85it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.86it/s] 84%|████████▍ | 27/32 [00:06<00:01, 3.86it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.86it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.85it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.85it/s] 97%|█████████▋| 31/32 [00:07<00:00, 3.85it/s] 100%|██████████| 32/32 [00:08<00:00, 3.85it/s] 100%|██████████| 32/32 [00:08<00:00, 3.88it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDb61azr59wsrma0cp9tfaa83qt0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.15
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.15, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.15' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:16:04.390683Z", "created_at": "2025-04-19T15:15:43.206000Z", "data_removed": false, "error": null, "id": "b61azr59wsrma0cp9tfaa83qt0", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "free=27118267809792\nDownloading weights\n2025-04-19T15:15:47Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpro06079u/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T15:15:47Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T15:15:55Z | INFO | [ Complete ] dest=/tmp/tmpro06079u/weights size=\"172 MB\" total_elapsed=7.748s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 7.77s\nLoaded LoRAs in 8.37s\nUsing seed: 14682\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.73it/s]\n 6%|▋ | 2/32 [00:00<00:07, 4.11it/s]\n 9%|▉ | 3/32 [00:00<00:07, 3.99it/s]\n 12%|█▎ | 4/32 [00:01<00:07, 3.93it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.90it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.88it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.87it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.86it/s]\n 28%|██▊ | 9/32 [00:02<00:05, 3.85it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.85it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.84it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.84it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.84it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.84it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.83it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.78it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.78it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.80it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.82it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.80it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.79it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.81it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.85it/s]\n 84%|████████▍ | 27/32 [00:07<00:01, 3.85it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.85it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.84it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.83it/s]\n 97%|█████████▋| 31/32 [00:08<00:00, 3.84it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.83it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 16.982306311, "total_time": 21.184683 }, "output": [ "https://replicate.delivery/xezq/NeI3EukqbbSjIy1eQeLDK1aEhfrTdsLlbYNy4PniuUGRDcRSB/out-0.webp" ], "started_at": "2025-04-19T15:15:47.408376Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-6h573fhd44zmyi2xvosx4ct6mh4yskwjj3svbhmzle6iojzzz7wq", "get": "https://api.replicate.com/v1/predictions/b61azr59wsrma0cp9tfaa83qt0", "cancel": "https://api.replicate.com/v1/predictions/b61azr59wsrma0cp9tfaa83qt0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=27118267809792 Downloading weights 2025-04-19T15:15:47Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpro06079u/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T15:15:47Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T15:15:55Z | INFO | [ Complete ] dest=/tmp/tmpro06079u/weights size="172 MB" total_elapsed=7.748s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 7.77s Loaded LoRAs in 8.37s Using seed: 14682 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.73it/s] 6%|▋ | 2/32 [00:00<00:07, 4.11it/s] 9%|▉ | 3/32 [00:00<00:07, 3.99it/s] 12%|█▎ | 4/32 [00:01<00:07, 3.93it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.90it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.88it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.87it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.86it/s] 28%|██▊ | 9/32 [00:02<00:05, 3.85it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.85it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.84it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.84it/s] 41%|████ | 13/32 [00:03<00:04, 3.84it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.84it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.83it/s] 50%|█████ | 16/32 [00:04<00:04, 3.78it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.78it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.80it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.82it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.80it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.79it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.81it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.85it/s] 84%|████████▍ | 27/32 [00:07<00:01, 3.85it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.85it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.84it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.83it/s] 97%|█████████▋| 31/32 [00:08<00:00, 3.84it/s] 100%|██████████| 32/32 [00:08<00:00, 3.83it/s] 100%|██████████| 32/32 [00:08<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDrxzm3r1gbsrmc0cp9tfrztbewcStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.15
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.15, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.15' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:16:30.471394Z", "created_at": "2025-04-19T15:16:17.630000Z", "data_removed": false, "error": null, "id": "rxzm3r1gbsrmc0cp9tfrztbewc", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.15, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "free=25969939996672\nDownloading weights\n2025-04-19T15:16:17Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp0e1khxmz/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T15:16:17Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T15:16:21Z | INFO | [ Complete ] dest=/tmp/tmp0e1khxmz/weights size=\"172 MB\" total_elapsed=3.399s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 3.42s\nLoaded LoRAs in 3.98s\nUsing seed: 45309\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.76it/s]\n 6%|▋ | 2/32 [00:00<00:07, 4.16it/s]\n 9%|▉ | 3/32 [00:00<00:07, 4.01it/s]\n 12%|█▎ | 4/32 [00:01<00:07, 3.95it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.90it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.88it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.87it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.86it/s]\n 28%|██▊ | 9/32 [00:02<00:05, 3.85it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.85it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.85it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.85it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.84it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.84it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.85it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.84it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.84it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.83it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.84it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.84it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.84it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.83it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.84it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.84it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.84it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.83it/s]\n 84%|████████▍ | 27/32 [00:07<00:01, 3.83it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.84it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.84it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.84it/s]\n 97%|█████████▋| 31/32 [00:08<00:00, 3.84it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.85it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.85it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 12.567578043, "total_time": 12.841394 }, "output": [ "https://replicate.delivery/xezq/2vioZURL3BpPDNi2kexSdRQ5nVjFpflgba646vsxm5KOBXkUA/out-0.webp" ], "started_at": "2025-04-19T15:16:17.903816Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-gci2md3mmj5ai6hb2wqtf3ngozpmm24yqx62vo2ygtrdfpp5czpq", "get": "https://api.replicate.com/v1/predictions/rxzm3r1gbsrmc0cp9tfrztbewc", "cancel": "https://api.replicate.com/v1/predictions/rxzm3r1gbsrmc0cp9tfrztbewc/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25969939996672 Downloading weights 2025-04-19T15:16:17Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp0e1khxmz/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T15:16:17Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T15:16:21Z | INFO | [ Complete ] dest=/tmp/tmp0e1khxmz/weights size="172 MB" total_elapsed=3.399s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 3.42s Loaded LoRAs in 3.98s Using seed: 45309 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.76it/s] 6%|▋ | 2/32 [00:00<00:07, 4.16it/s] 9%|▉ | 3/32 [00:00<00:07, 4.01it/s] 12%|█▎ | 4/32 [00:01<00:07, 3.95it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.90it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.88it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.87it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.86it/s] 28%|██▊ | 9/32 [00:02<00:05, 3.85it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.85it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.85it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.85it/s] 41%|████ | 13/32 [00:03<00:04, 3.84it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.84it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.85it/s] 50%|█████ | 16/32 [00:04<00:04, 3.84it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.84it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.83it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.84it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.84it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.84it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.83it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.84it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.84it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.84it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.83it/s] 84%|████████▍ | 27/32 [00:07<00:01, 3.83it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.84it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.84it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.84it/s] 97%|█████████▋| 31/32 [00:08<00:00, 3.84it/s] 100%|██████████| 32/32 [00:08<00:00, 3.85it/s] 100%|██████████| 32/32 [00:08<00:00, 3.85it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID43qg5z5nf5rmc0cp9tfvwrarr0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 3
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 3, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=3' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:17:00.964073Z", "created_at": "2025-04-19T15:16:51.705000Z", "data_removed": false, "error": null, "id": "43qg5z5nf5rmc0cp9tfvwrarr0", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 27877\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.73it/s]\n 6%|▋ | 2/32 [00:00<00:07, 4.14it/s]\n 9%|▉ | 3/32 [00:00<00:07, 3.98it/s]\n 12%|█▎ | 4/32 [00:01<00:07, 3.92it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.88it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.86it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.85it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.85it/s]\n 28%|██▊ | 9/32 [00:02<00:05, 3.84it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.83it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.82it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.82it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.82it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.82it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.81it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.81it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.82it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.82it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.82it/s]\n 84%|████████▍ | 27/32 [00:07<00:01, 3.81it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.81it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.82it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.81it/s]\n 97%|█████████▋| 31/32 [00:08<00:00, 3.82it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.82it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.190636015, "total_time": 9.259073 }, "output": [ "https://replicate.delivery/xezq/mfyyUZLB1G3cZieMrKpuXxMuuhhB9TWsK3orfz1OoATYDuIpA/out-0.webp" ], "started_at": "2025-04-19T15:16:51.773437Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-bjnztguqzullepl4ue6fvmasygzio5sen5zsawts2p4htycai33a", "get": "https://api.replicate.com/v1/predictions/43qg5z5nf5rmc0cp9tfvwrarr0", "cancel": "https://api.replicate.com/v1/predictions/43qg5z5nf5rmc0cp9tfvwrarr0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 27877 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.73it/s] 6%|▋ | 2/32 [00:00<00:07, 4.14it/s] 9%|▉ | 3/32 [00:00<00:07, 3.98it/s] 12%|█▎ | 4/32 [00:01<00:07, 3.92it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.88it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.86it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.85it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.85it/s] 28%|██▊ | 9/32 [00:02<00:05, 3.84it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.83it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.82it/s] 41%|████ | 13/32 [00:03<00:04, 3.82it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s] 50%|█████ | 16/32 [00:04<00:04, 3.82it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.82it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.82it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.81it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.81it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.82it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.82it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.82it/s] 84%|████████▍ | 27/32 [00:07<00:01, 3.81it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.81it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.82it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.81it/s] 97%|█████████▋| 31/32 [00:08<00:00, 3.82it/s] 100%|██████████| 32/32 [00:08<00:00, 3.82it/s] 100%|██████████| 32/32 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDjnwtzq1ce9rme0cp9tgb19g0qcStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 3
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 3, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=3' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:17:32.922992Z", "created_at": "2025-04-19T15:17:22.162000Z", "data_removed": false, "error": null, "id": "jnwtzq1ce9rme0cp9tgb19g0qc", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 46480\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/38 [00:00<?, ?it/s]\n 3%|▎ | 1/38 [00:00<00:09, 3.73it/s]\n 5%|▌ | 2/38 [00:00<00:08, 4.14it/s]\n 8%|▊ | 3/38 [00:00<00:08, 3.98it/s]\n 11%|█ | 4/38 [00:01<00:08, 3.92it/s]\n 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s]\n 16%|█▌ | 6/38 [00:01<00:08, 3.86it/s]\n 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s]\n 21%|██ | 8/38 [00:02<00:07, 3.84it/s]\n 24%|██▎ | 9/38 [00:02<00:07, 3.84it/s]\n 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s]\n 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s]\n 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s]\n 34%|███▍ | 13/38 [00:03<00:06, 3.83it/s]\n 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s]\n 39%|███▉ | 15/38 [00:03<00:06, 3.82it/s]\n 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s]\n 45%|████▍ | 17/38 [00:04<00:05, 3.82it/s]\n 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s]\n 50%|█████ | 19/38 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 20/38 [00:05<00:04, 3.83it/s]\n 55%|█████▌ | 21/38 [00:05<00:04, 3.83it/s]\n 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s]\n 61%|██████ | 23/38 [00:05<00:03, 3.82it/s]\n 63%|██████▎ | 24/38 [00:06<00:03, 3.83it/s]\n 66%|██████▌ | 25/38 [00:06<00:03, 3.83it/s]\n 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s]\n 71%|███████ | 27/38 [00:07<00:02, 3.82it/s]\n 74%|███████▎ | 28/38 [00:07<00:02, 3.83it/s]\n 76%|███████▋ | 29/38 [00:07<00:02, 3.83it/s]\n 79%|███████▉ | 30/38 [00:07<00:02, 3.82it/s]\n 82%|████████▏ | 31/38 [00:08<00:01, 3.83it/s]\n 84%|████████▍ | 32/38 [00:08<00:01, 3.83it/s]\n 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s]\n 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s]\n 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s]\n 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s]\n 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.81it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.753560609, "total_time": 10.760992 }, "output": [ "https://replicate.delivery/xezq/eX2ng8gNPcWMfErCCviPIzBiOQbxIUWlyC5xO0EFNcWMCXkUA/out-0.webp" ], "started_at": "2025-04-19T15:17:22.169431Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-lhmsxqrede7qvqoawqwo6s354khatv7aynaq3cr4zbzkqxtzhota", "get": "https://api.replicate.com/v1/predictions/jnwtzq1ce9rme0cp9tgb19g0qc", "cancel": "https://api.replicate.com/v1/predictions/jnwtzq1ce9rme0cp9tgb19g0qc/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 46480 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/38 [00:00<?, ?it/s] 3%|▎ | 1/38 [00:00<00:09, 3.73it/s] 5%|▌ | 2/38 [00:00<00:08, 4.14it/s] 8%|▊ | 3/38 [00:00<00:08, 3.98it/s] 11%|█ | 4/38 [00:01<00:08, 3.92it/s] 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s] 16%|█▌ | 6/38 [00:01<00:08, 3.86it/s] 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s] 21%|██ | 8/38 [00:02<00:07, 3.84it/s] 24%|██▎ | 9/38 [00:02<00:07, 3.84it/s] 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s] 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s] 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s] 34%|███▍ | 13/38 [00:03<00:06, 3.83it/s] 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s] 39%|███▉ | 15/38 [00:03<00:06, 3.82it/s] 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s] 45%|████▍ | 17/38 [00:04<00:05, 3.82it/s] 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s] 50%|█████ | 19/38 [00:04<00:04, 3.82it/s] 53%|█████▎ | 20/38 [00:05<00:04, 3.83it/s] 55%|█████▌ | 21/38 [00:05<00:04, 3.83it/s] 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s] 61%|██████ | 23/38 [00:05<00:03, 3.82it/s] 63%|██████▎ | 24/38 [00:06<00:03, 3.83it/s] 66%|██████▌ | 25/38 [00:06<00:03, 3.83it/s] 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s] 71%|███████ | 27/38 [00:07<00:02, 3.82it/s] 74%|███████▎ | 28/38 [00:07<00:02, 3.83it/s] 76%|███████▋ | 29/38 [00:07<00:02, 3.83it/s] 79%|███████▉ | 30/38 [00:07<00:02, 3.82it/s] 82%|████████▏ | 31/38 [00:08<00:01, 3.83it/s] 84%|████████▍ | 32/38 [00:08<00:01, 3.83it/s] 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s] 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s] 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s] 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s] 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s] 100%|██████████| 38/38 [00:09<00:00, 3.81it/s] 100%|██████████| 38/38 [00:09<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDb3h8s4vwp1rmc0cp9tgbnah51mStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 5.3
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 5.3, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=5.3' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:17:56.444057Z", "created_at": "2025-04-19T15:17:42.704000Z", "data_removed": false, "error": null, "id": "b3h8s4vwp1rmc0cp9tgbnah51m", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 2.65s\nUsing seed: 27119\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/38 [00:00<?, ?it/s]\n 3%|▎ | 1/38 [00:00<00:09, 3.75it/s]\n 5%|▌ | 2/38 [00:00<00:08, 4.15it/s]\n 8%|▊ | 3/38 [00:00<00:08, 4.01it/s]\n 11%|█ | 4/38 [00:01<00:08, 3.93it/s]\n 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s]\n 16%|█▌ | 6/38 [00:01<00:08, 3.86it/s]\n 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s]\n 21%|██ | 8/38 [00:02<00:07, 3.84it/s]\n 24%|██▎ | 9/38 [00:02<00:07, 3.83it/s]\n 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s]\n 29%|██▉ | 11/38 [00:02<00:07, 3.84it/s]\n 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s]\n 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s]\n 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s]\n 39%|███▉ | 15/38 [00:03<00:06, 3.83it/s]\n 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s]\n 45%|████▍ | 17/38 [00:04<00:05, 3.82it/s]\n 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s]\n 50%|█████ | 19/38 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 20/38 [00:05<00:04, 3.81it/s]\n 55%|█████▌ | 21/38 [00:05<00:04, 3.82it/s]\n 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s]\n 61%|██████ | 23/38 [00:05<00:03, 3.81it/s]\n 63%|██████▎ | 24/38 [00:06<00:03, 3.82it/s]\n 66%|██████▌ | 25/38 [00:06<00:03, 3.82it/s]\n 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s]\n 71%|███████ | 27/38 [00:07<00:02, 3.82it/s]\n 74%|███████▎ | 28/38 [00:07<00:02, 3.82it/s]\n 76%|███████▋ | 29/38 [00:07<00:02, 3.82it/s]\n 79%|███████▉ | 30/38 [00:07<00:02, 3.83it/s]\n 82%|████████▏ | 31/38 [00:08<00:01, 3.82it/s]\n 84%|████████▍ | 32/38 [00:08<00:01, 3.82it/s]\n 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s]\n 89%|████████▉ | 34/38 [00:08<00:01, 3.83it/s]\n 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s]\n 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s]\n 97%|█████████▋| 37/38 [00:09<00:00, 3.82it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.83it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 12.844170971, "total_time": 13.740057 }, "output": [ "https://replicate.delivery/xezq/wekcnhFRyegXNUnr55keTBDVqL0qXQMC4sonICNpUvTIFuIpA/out-0.webp" ], "started_at": "2025-04-19T15:17:43.599886Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-lhmz2qjpfahiylqrdpxuv3bvijloqpxwpffujesgziutevy6dcua", "get": "https://api.replicate.com/v1/predictions/b3h8s4vwp1rmc0cp9tgbnah51m", "cancel": "https://api.replicate.com/v1/predictions/b3h8s4vwp1rmc0cp9tgbnah51m/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 2.65s Using seed: 27119 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/38 [00:00<?, ?it/s] 3%|▎ | 1/38 [00:00<00:09, 3.75it/s] 5%|▌ | 2/38 [00:00<00:08, 4.15it/s] 8%|▊ | 3/38 [00:00<00:08, 4.01it/s] 11%|█ | 4/38 [00:01<00:08, 3.93it/s] 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s] 16%|█▌ | 6/38 [00:01<00:08, 3.86it/s] 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s] 21%|██ | 8/38 [00:02<00:07, 3.84it/s] 24%|██▎ | 9/38 [00:02<00:07, 3.83it/s] 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s] 29%|██▉ | 11/38 [00:02<00:07, 3.84it/s] 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s] 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s] 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s] 39%|███▉ | 15/38 [00:03<00:06, 3.83it/s] 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s] 45%|████▍ | 17/38 [00:04<00:05, 3.82it/s] 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s] 50%|█████ | 19/38 [00:04<00:04, 3.82it/s] 53%|█████▎ | 20/38 [00:05<00:04, 3.81it/s] 55%|█████▌ | 21/38 [00:05<00:04, 3.82it/s] 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s] 61%|██████ | 23/38 [00:05<00:03, 3.81it/s] 63%|██████▎ | 24/38 [00:06<00:03, 3.82it/s] 66%|██████▌ | 25/38 [00:06<00:03, 3.82it/s] 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s] 71%|███████ | 27/38 [00:07<00:02, 3.82it/s] 74%|███████▎ | 28/38 [00:07<00:02, 3.82it/s] 76%|███████▋ | 29/38 [00:07<00:02, 3.82it/s] 79%|███████▉ | 30/38 [00:07<00:02, 3.83it/s] 82%|████████▏ | 31/38 [00:08<00:01, 3.82it/s] 84%|████████▍ | 32/38 [00:08<00:01, 3.82it/s] 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s] 89%|████████▉ | 34/38 [00:08<00:01, 3.83it/s] 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s] 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s] 97%|█████████▋| 37/38 [00:09<00:00, 3.82it/s] 100%|██████████| 38/38 [00:09<00:00, 3.83it/s] 100%|██████████| 38/38 [00:09<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID9tx72rg799rma0cp9tgtfqcak4StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 8.17
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.17, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 8.17, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.17, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.17, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=8.17' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \\nStyle: ethereal, dreamlike \\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \\nColor Palette: muted pastels—blush pinks, powder blues, warm golds \\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.17, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:18:31.075127Z", "created_at": "2025-04-19T15:18:18.186000Z", "data_removed": false, "error": null, "id": "9tx72rg799rma0cp9tgtfqcak4", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head \nStyle: ethereal, dreamlike \nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges \nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt \nColor Palette: muted pastels—blush pinks, powder blues, warm golds \nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy \nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds \nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.17, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 0.58s\nUsing seed: 3503\nPrompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head\nStyle: ethereal, dreamlike\nComposition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges\nLighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt\nColor Palette: muted pastels—blush pinks, powder blues, warm golds\nMood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy\nTechnical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds\nAdditional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\n[!] txt2img mode\n 0%| | 0/38 [00:00<?, ?it/s]\n 3%|▎ | 1/38 [00:00<00:09, 3.72it/s]\n 5%|▌ | 2/38 [00:00<00:08, 4.12it/s]\n 8%|▊ | 3/38 [00:00<00:08, 3.98it/s]\n 11%|█ | 4/38 [00:01<00:08, 3.92it/s]\n 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s]\n 16%|█▌ | 6/38 [00:01<00:08, 3.80it/s]\n 18%|█▊ | 7/38 [00:01<00:08, 3.82it/s]\n 21%|██ | 8/38 [00:02<00:07, 3.85it/s]\n 24%|██▎ | 9/38 [00:02<00:07, 3.80it/s]\n 26%|██▋ | 10/38 [00:02<00:07, 3.82it/s]\n 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s]\n 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s]\n 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s]\n 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s]\n 39%|███▉ | 15/38 [00:03<00:05, 3.87it/s]\n 42%|████▏ | 16/38 [00:04<00:05, 3.86it/s]\n 45%|████▍ | 17/38 [00:04<00:05, 3.85it/s]\n 47%|████▋ | 18/38 [00:04<00:05, 3.84it/s]\n 50%|█████ | 19/38 [00:04<00:04, 3.84it/s]\n 53%|█████▎ | 20/38 [00:05<00:04, 3.84it/s]\n 55%|█████▌ | 21/38 [00:05<00:04, 3.84it/s]\n 58%|█████▊ | 22/38 [00:05<00:04, 3.83it/s]\n 61%|██████ | 23/38 [00:05<00:03, 3.83it/s]\n 63%|██████▎ | 24/38 [00:06<00:03, 3.83it/s]\n 66%|██████▌ | 25/38 [00:06<00:03, 3.83it/s]\n 68%|██████▊ | 26/38 [00:06<00:03, 3.83it/s]\n 71%|███████ | 27/38 [00:07<00:02, 3.82it/s]\n 74%|███████▎ | 28/38 [00:07<00:02, 3.77it/s]\n 76%|███████▋ | 29/38 [00:07<00:02, 3.80it/s]\n 79%|███████▉ | 30/38 [00:07<00:02, 3.80it/s]\n 82%|████████▏ | 31/38 [00:08<00:01, 3.80it/s]\n 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s]\n 87%|████████▋ | 33/38 [00:08<00:01, 3.80it/s]\n 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s]\n 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s]\n 95%|█████████▍| 36/38 [00:09<00:00, 3.81it/s]\n 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.82it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.808184902, "total_time": 12.889127 }, "output": [ "https://replicate.delivery/xezq/LvmevCYfSwsOYEURJciGEvrLoAI0viPyfUteBEVoywYdMcRSB/out-0.webp" ], "started_at": "2025-04-19T15:18:20.266942Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-wgq6dawtv4qyozbvxqdk6can36uis745zyz3o4eickx7jbawetoq", "get": "https://api.replicate.com/v1/predictions/9tx72rg799rma0cp9tgtfqcak4", "cancel": "https://api.replicate.com/v1/predictions/9tx72rg799rma0cp9tgtfqcak4/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.58s Using seed: 3503 Prompt: Subject: TOKMAHI’s full face and neck clearly visible, filling most of the frame, sporting an exaggerated, gravity‑defying afro sculpted into swirling cloud‑like tendrils that arc wildly above her head Style: ethereal, dreamlike Composition: centered, tight crop from shoulders up, inverted gravity hinted by soft pastel clouds at the edges Lighting: soft golden shafts highlighting her facial contours, subtle rim light defining the dramatic hair sculpt Color Palette: muted pastels—blush pinks, powder blues, warm golds Mood/Atmosphere: serene, weightless, intimate with a touch of whimsical fantasy Technical Details: tack‑sharp focus on eyes, lips, and neckline; gentle motion blur on hair tips and surrounding clouds Additional Elements: faint halos of upside‑down rainbows and drifting lanterns just beyond the crop --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 [!] txt2img mode 0%| | 0/38 [00:00<?, ?it/s] 3%|▎ | 1/38 [00:00<00:09, 3.72it/s] 5%|▌ | 2/38 [00:00<00:08, 4.12it/s] 8%|▊ | 3/38 [00:00<00:08, 3.98it/s] 11%|█ | 4/38 [00:01<00:08, 3.92it/s] 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s] 16%|█▌ | 6/38 [00:01<00:08, 3.80it/s] 18%|█▊ | 7/38 [00:01<00:08, 3.82it/s] 21%|██ | 8/38 [00:02<00:07, 3.85it/s] 24%|██▎ | 9/38 [00:02<00:07, 3.80it/s] 26%|██▋ | 10/38 [00:02<00:07, 3.82it/s] 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s] 32%|███▏ | 12/38 [00:03<00:06, 3.83it/s] 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s] 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s] 39%|███▉ | 15/38 [00:03<00:05, 3.87it/s] 42%|████▏ | 16/38 [00:04<00:05, 3.86it/s] 45%|████▍ | 17/38 [00:04<00:05, 3.85it/s] 47%|████▋ | 18/38 [00:04<00:05, 3.84it/s] 50%|█████ | 19/38 [00:04<00:04, 3.84it/s] 53%|█████▎ | 20/38 [00:05<00:04, 3.84it/s] 55%|█████▌ | 21/38 [00:05<00:04, 3.84it/s] 58%|█████▊ | 22/38 [00:05<00:04, 3.83it/s] 61%|██████ | 23/38 [00:05<00:03, 3.83it/s] 63%|██████▎ | 24/38 [00:06<00:03, 3.83it/s] 66%|██████▌ | 25/38 [00:06<00:03, 3.83it/s] 68%|██████▊ | 26/38 [00:06<00:03, 3.83it/s] 71%|███████ | 27/38 [00:07<00:02, 3.82it/s] 74%|███████▎ | 28/38 [00:07<00:02, 3.77it/s] 76%|███████▋ | 29/38 [00:07<00:02, 3.80it/s] 79%|███████▉ | 30/38 [00:07<00:02, 3.80it/s] 82%|████████▏ | 31/38 [00:08<00:01, 3.80it/s] 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s] 87%|████████▋ | 33/38 [00:08<00:01, 3.80it/s] 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s] 92%|█████████▏| 35/38 [00:09<00:00, 3.82it/s] 95%|█████████▍| 36/38 [00:09<00:00, 3.81it/s] 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s] 100%|██████████| 38/38 [00:09<00:00, 3.82it/s] 100%|██████████| 38/38 [00:09<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDcb9q418zb5rma0cp9tk8dn8vv8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 5
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 5, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=5' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:24:01.489089Z", "created_at": "2025-04-19T15:23:52.025000Z", "data_removed": false, "error": null, "id": "cb9q418zb5rma0cp9tk8dn8vv8", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 8444\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.74it/s]\n 6%|▋ | 2/32 [00:00<00:07, 4.13it/s]\n 9%|▉ | 3/32 [00:00<00:07, 3.98it/s]\n 12%|█▎ | 4/32 [00:01<00:07, 3.92it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.88it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.87it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.84it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.83it/s]\n 28%|██▊ | 9/32 [00:02<00:06, 3.83it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.83it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.82it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.82it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.82it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.81it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.81it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.81it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.81it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.81it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.82it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.81it/s]\n 84%|████████▍ | 27/32 [00:07<00:01, 3.82it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.82it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.82it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.81it/s]\n 97%|█████████▋| 31/32 [00:08<00:00, 3.82it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.81it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.202414608, "total_time": 9.464089 }, "output": [ "https://replicate.delivery/xezq/C2QAEUFhP7rbJp33KI1XyCEm8Z3YN6x47mPiStSXOqXEyFJF/out-0.webp" ], "started_at": "2025-04-19T15:23:52.286674Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-7e4zrgkwlmfqpjf3ujeby7azemmjylkveyzbe55qso2dzhssogdq", "get": "https://api.replicate.com/v1/predictions/cb9q418zb5rma0cp9tk8dn8vv8", "cancel": "https://api.replicate.com/v1/predictions/cb9q418zb5rma0cp9tk8dn8vv8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 8444 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.74it/s] 6%|▋ | 2/32 [00:00<00:07, 4.13it/s] 9%|▉ | 3/32 [00:00<00:07, 3.98it/s] 12%|█▎ | 4/32 [00:01<00:07, 3.92it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.88it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.87it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.84it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.83it/s] 28%|██▊ | 9/32 [00:02<00:06, 3.83it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.83it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.82it/s] 41%|████ | 13/32 [00:03<00:04, 3.82it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s] 50%|█████ | 16/32 [00:04<00:04, 3.82it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.82it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.81it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.81it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.81it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.81it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.81it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.82it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.81it/s] 84%|████████▍ | 27/32 [00:07<00:01, 3.82it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.82it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.82it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.81it/s] 97%|█████████▋| 31/32 [00:08<00:00, 3.82it/s] 100%|██████████| 32/32 [00:08<00:00, 3.81it/s] 100%|██████████| 32/32 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID8rq4kagph5rmc0cp9tkt3khnhwStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 32
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 32 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=32'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:25:05.199442Z", "created_at": "2025-04-19T15:24:55.305000Z", "data_removed": false, "error": null, "id": "8rq4kagph5rmc0cp9tkt3khnhw", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 32 }, "logs": "free=25601314668544\nDownloading weights\n2025-04-19T15:24:55Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpg93dmgni/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T15:24:55Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T15:24:56Z | INFO | [ Complete ] dest=/tmp/tmpg93dmgni/weights size=\"172 MB\" total_elapsed=0.543s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 0.56s\nLoaded LoRAs in 1.12s\nUsing seed: 492\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/32 [00:00<?, ?it/s]\n 3%|▎ | 1/32 [00:00<00:08, 3.72it/s]\n 6%|▋ | 2/32 [00:00<00:07, 4.11it/s]\n 9%|▉ | 3/32 [00:00<00:07, 3.99it/s]\n 12%|█▎ | 4/32 [00:01<00:07, 3.91it/s]\n 16%|█▌ | 5/32 [00:01<00:06, 3.87it/s]\n 19%|█▉ | 6/32 [00:01<00:06, 3.85it/s]\n 22%|██▏ | 7/32 [00:01<00:06, 3.84it/s]\n 25%|██▌ | 8/32 [00:02<00:06, 3.83it/s]\n 28%|██▊ | 9/32 [00:02<00:06, 3.82it/s]\n 31%|███▏ | 10/32 [00:02<00:05, 3.82it/s]\n 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s]\n 38%|███▊ | 12/32 [00:03<00:05, 3.81it/s]\n 41%|████ | 13/32 [00:03<00:04, 3.81it/s]\n 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s]\n 50%|█████ | 16/32 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 17/32 [00:04<00:03, 3.81it/s]\n 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s]\n 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s]\n 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s]\n 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s]\n 69%|██████▉ | 22/32 [00:05<00:02, 3.82it/s]\n 72%|███████▏ | 23/32 [00:05<00:02, 3.82it/s]\n 75%|███████▌ | 24/32 [00:06<00:02, 3.81it/s]\n 78%|███████▊ | 25/32 [00:06<00:01, 3.81it/s]\n 81%|████████▏ | 26/32 [00:06<00:01, 3.82it/s]\n 84%|████████▍ | 27/32 [00:07<00:01, 3.82it/s]\n 88%|████████▊ | 28/32 [00:07<00:01, 3.82it/s]\n 91%|█████████ | 29/32 [00:07<00:00, 3.81it/s]\n 94%|█████████▍| 30/32 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 31/32 [00:08<00:00, 3.81it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.81it/s]\n100%|██████████| 32/32 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.771369431, "total_time": 9.894442 }, "output": [ "https://replicate.delivery/xezq/3teqgSkFpTVVRKbk9P8vGkHjM9fQWQSUpF1akeWRtWTiSuIpA/out-0.webp" ], "started_at": "2025-04-19T15:24:55.428073Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-ctngyj3julvqvtjydbp3eheitzfw7ginumwx3hxnd5tmxvnrfyga", "get": "https://api.replicate.com/v1/predictions/8rq4kagph5rmc0cp9tkt3khnhw", "cancel": "https://api.replicate.com/v1/predictions/8rq4kagph5rmc0cp9tkt3khnhw/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25601314668544 Downloading weights 2025-04-19T15:24:55Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpg93dmgni/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T15:24:55Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T15:24:56Z | INFO | [ Complete ] dest=/tmp/tmpg93dmgni/weights size="172 MB" total_elapsed=0.543s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 0.56s Loaded LoRAs in 1.12s Using seed: 492 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/32 [00:00<?, ?it/s] 3%|▎ | 1/32 [00:00<00:08, 3.72it/s] 6%|▋ | 2/32 [00:00<00:07, 4.11it/s] 9%|▉ | 3/32 [00:00<00:07, 3.99it/s] 12%|█▎ | 4/32 [00:01<00:07, 3.91it/s] 16%|█▌ | 5/32 [00:01<00:06, 3.87it/s] 19%|█▉ | 6/32 [00:01<00:06, 3.85it/s] 22%|██▏ | 7/32 [00:01<00:06, 3.84it/s] 25%|██▌ | 8/32 [00:02<00:06, 3.83it/s] 28%|██▊ | 9/32 [00:02<00:06, 3.82it/s] 31%|███▏ | 10/32 [00:02<00:05, 3.82it/s] 34%|███▍ | 11/32 [00:02<00:05, 3.82it/s] 38%|███▊ | 12/32 [00:03<00:05, 3.81it/s] 41%|████ | 13/32 [00:03<00:04, 3.81it/s] 44%|████▍ | 14/32 [00:03<00:04, 3.82it/s] 47%|████▋ | 15/32 [00:03<00:04, 3.82it/s] 50%|█████ | 16/32 [00:04<00:04, 3.82it/s] 53%|█████▎ | 17/32 [00:04<00:03, 3.81it/s] 56%|█████▋ | 18/32 [00:04<00:03, 3.82it/s] 59%|█████▉ | 19/32 [00:04<00:03, 3.82it/s] 62%|██████▎ | 20/32 [00:05<00:03, 3.82it/s] 66%|██████▌ | 21/32 [00:05<00:02, 3.81it/s] 69%|██████▉ | 22/32 [00:05<00:02, 3.82it/s] 72%|███████▏ | 23/32 [00:05<00:02, 3.82it/s] 75%|███████▌ | 24/32 [00:06<00:02, 3.81it/s] 78%|███████▊ | 25/32 [00:06<00:01, 3.81it/s] 81%|████████▏ | 26/32 [00:06<00:01, 3.82it/s] 84%|████████▍ | 27/32 [00:07<00:01, 3.82it/s] 88%|████████▊ | 28/32 [00:07<00:01, 3.82it/s] 91%|█████████ | 29/32 [00:07<00:00, 3.81it/s] 94%|█████████▍| 30/32 [00:07<00:00, 3.82it/s] 97%|█████████▋| 31/32 [00:08<00:00, 3.81it/s] 100%|██████████| 32/32 [00:08<00:00, 3.81it/s] 100%|██████████| 32/32 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDtb0edvmj4hrma0cp9tkt95mwpmStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 44
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 44 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=44'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:25:41.712397Z", "created_at": "2025-04-19T15:25:26.948000Z", "data_removed": false, "error": null, "id": "tb0edvmj4hrma0cp9tkt95mwpm", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 }, "logs": "free=27318501900288\nDownloading weights\n2025-04-19T15:25:27Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpk7z7dn9_/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T15:25:27Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T15:25:29Z | INFO | [ Complete ] dest=/tmp/tmpk7z7dn9_/weights size=\"172 MB\" total_elapsed=2.324s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 2.34s\nLoaded LoRAs in 2.91s\nUsing seed: 61226\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/44 [00:00<?, ?it/s]\n 2%|▏ | 1/44 [00:00<00:11, 3.73it/s]\n 5%|▍ | 2/44 [00:00<00:10, 4.12it/s]\n 7%|▋ | 3/44 [00:00<00:10, 3.97it/s]\n 9%|▉ | 4/44 [00:01<00:10, 3.90it/s]\n 11%|█▏ | 5/44 [00:01<00:10, 3.85it/s]\n 14%|█▎ | 6/44 [00:01<00:09, 3.83it/s]\n 16%|█▌ | 7/44 [00:01<00:09, 3.82it/s]\n 18%|█▊ | 8/44 [00:02<00:09, 3.82it/s]\n 20%|██ | 9/44 [00:02<00:09, 3.82it/s]\n 23%|██▎ | 10/44 [00:02<00:08, 3.81it/s]\n 25%|██▌ | 11/44 [00:02<00:08, 3.81it/s]\n 27%|██▋ | 12/44 [00:03<00:08, 3.80it/s]\n 30%|██▉ | 13/44 [00:03<00:08, 3.81it/s]\n 32%|███▏ | 14/44 [00:03<00:07, 3.80it/s]\n 34%|███▍ | 15/44 [00:03<00:07, 3.80it/s]\n 36%|███▋ | 16/44 [00:04<00:07, 3.80it/s]\n 39%|███▊ | 17/44 [00:04<00:07, 3.80it/s]\n 41%|████ | 18/44 [00:04<00:06, 3.79it/s]\n 43%|████▎ | 19/44 [00:04<00:06, 3.80it/s]\n 45%|████▌ | 20/44 [00:05<00:06, 3.80it/s]\n 48%|████▊ | 21/44 [00:05<00:06, 3.80it/s]\n 50%|█████ | 22/44 [00:05<00:05, 3.80it/s]\n 52%|█████▏ | 23/44 [00:06<00:05, 3.80it/s]\n 55%|█████▍ | 24/44 [00:06<00:05, 3.80it/s]\n 57%|█████▋ | 25/44 [00:06<00:04, 3.80it/s]\n 59%|█████▉ | 26/44 [00:06<00:04, 3.80it/s]\n 61%|██████▏ | 27/44 [00:07<00:04, 3.80it/s]\n 64%|██████▎ | 28/44 [00:07<00:04, 3.80it/s]\n 66%|██████▌ | 29/44 [00:07<00:03, 3.80it/s]\n 68%|██████▊ | 30/44 [00:07<00:03, 3.80it/s]\n 70%|███████ | 31/44 [00:08<00:03, 3.79it/s]\n 73%|███████▎ | 32/44 [00:08<00:03, 3.80it/s]\n 75%|███████▌ | 33/44 [00:08<00:02, 3.80it/s]\n 77%|███████▋ | 34/44 [00:08<00:02, 3.79it/s]\n 80%|███████▉ | 35/44 [00:09<00:02, 3.79it/s]\n 82%|████████▏ | 36/44 [00:09<00:02, 3.80it/s]\n 84%|████████▍ | 37/44 [00:09<00:01, 3.80it/s]\n 86%|████████▋ | 38/44 [00:09<00:01, 3.80it/s]\n 89%|████████▊ | 39/44 [00:10<00:01, 3.79it/s]\n 91%|█████████ | 40/44 [00:10<00:01, 3.81it/s]\n 93%|█████████▎| 41/44 [00:10<00:00, 3.80it/s]\n 95%|█████████▌| 42/44 [00:11<00:00, 3.80it/s]\n 98%|█████████▊| 43/44 [00:11<00:00, 3.80it/s]\n100%|██████████| 44/44 [00:11<00:00, 3.79it/s]\n100%|██████████| 44/44 [00:11<00:00, 3.81it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 14.750104579, "total_time": 14.764397 }, "output": [ "https://replicate.delivery/xezq/3NNSitvUJ3KNOtbu1LFYac54fp9lUonUtBTvgpO3Tm16kLSKA/out-0.webp" ], "started_at": "2025-04-19T15:25:26.962293Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-hn4muerzlprltul26tl2dtrv7uamwuq2lm652z45lmizgkep25cq", "get": "https://api.replicate.com/v1/predictions/tb0edvmj4hrma0cp9tkt95mwpm", "cancel": "https://api.replicate.com/v1/predictions/tb0edvmj4hrma0cp9tkt95mwpm/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=27318501900288 Downloading weights 2025-04-19T15:25:27Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpk7z7dn9_/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T15:25:27Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T15:25:29Z | INFO | [ Complete ] dest=/tmp/tmpk7z7dn9_/weights size="172 MB" total_elapsed=2.324s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 2.34s Loaded LoRAs in 2.91s Using seed: 61226 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/44 [00:00<?, ?it/s] 2%|▏ | 1/44 [00:00<00:11, 3.73it/s] 5%|▍ | 2/44 [00:00<00:10, 4.12it/s] 7%|▋ | 3/44 [00:00<00:10, 3.97it/s] 9%|▉ | 4/44 [00:01<00:10, 3.90it/s] 11%|█▏ | 5/44 [00:01<00:10, 3.85it/s] 14%|█▎ | 6/44 [00:01<00:09, 3.83it/s] 16%|█▌ | 7/44 [00:01<00:09, 3.82it/s] 18%|█▊ | 8/44 [00:02<00:09, 3.82it/s] 20%|██ | 9/44 [00:02<00:09, 3.82it/s] 23%|██▎ | 10/44 [00:02<00:08, 3.81it/s] 25%|██▌ | 11/44 [00:02<00:08, 3.81it/s] 27%|██▋ | 12/44 [00:03<00:08, 3.80it/s] 30%|██▉ | 13/44 [00:03<00:08, 3.81it/s] 32%|███▏ | 14/44 [00:03<00:07, 3.80it/s] 34%|███▍ | 15/44 [00:03<00:07, 3.80it/s] 36%|███▋ | 16/44 [00:04<00:07, 3.80it/s] 39%|███▊ | 17/44 [00:04<00:07, 3.80it/s] 41%|████ | 18/44 [00:04<00:06, 3.79it/s] 43%|████▎ | 19/44 [00:04<00:06, 3.80it/s] 45%|████▌ | 20/44 [00:05<00:06, 3.80it/s] 48%|████▊ | 21/44 [00:05<00:06, 3.80it/s] 50%|█████ | 22/44 [00:05<00:05, 3.80it/s] 52%|█████▏ | 23/44 [00:06<00:05, 3.80it/s] 55%|█████▍ | 24/44 [00:06<00:05, 3.80it/s] 57%|█████▋ | 25/44 [00:06<00:04, 3.80it/s] 59%|█████▉ | 26/44 [00:06<00:04, 3.80it/s] 61%|██████▏ | 27/44 [00:07<00:04, 3.80it/s] 64%|██████▎ | 28/44 [00:07<00:04, 3.80it/s] 66%|██████▌ | 29/44 [00:07<00:03, 3.80it/s] 68%|██████▊ | 30/44 [00:07<00:03, 3.80it/s] 70%|███████ | 31/44 [00:08<00:03, 3.79it/s] 73%|███████▎ | 32/44 [00:08<00:03, 3.80it/s] 75%|███████▌ | 33/44 [00:08<00:02, 3.80it/s] 77%|███████▋ | 34/44 [00:08<00:02, 3.79it/s] 80%|███████▉ | 35/44 [00:09<00:02, 3.79it/s] 82%|████████▏ | 36/44 [00:09<00:02, 3.80it/s] 84%|████████▍ | 37/44 [00:09<00:01, 3.80it/s] 86%|████████▋ | 38/44 [00:09<00:01, 3.80it/s] 89%|████████▊ | 39/44 [00:10<00:01, 3.79it/s] 91%|█████████ | 40/44 [00:10<00:01, 3.81it/s] 93%|█████████▎| 41/44 [00:10<00:00, 3.80it/s] 95%|█████████▌| 42/44 [00:11<00:00, 3.80it/s] 98%|█████████▊| 43/44 [00:11<00:00, 3.80it/s] 100%|██████████| 44/44 [00:11<00:00, 3.79it/s] 100%|██████████| 44/44 [00:11<00:00, 3.81it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDrkc63xyc5xrma0cp9tmtmj534mStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 8.37
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.37, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 8.37, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.37, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.37, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=8.37' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.37, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:28:01.919513Z", "created_at": "2025-04-19T15:27:52.879000Z", "data_removed": false, "error": null, "id": "rkc63xyc5xrma0cp9tmtmj534m", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 8.37, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 1225\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.73it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.11it/s]\n 10%|█ | 3/30 [00:00<00:06, 3.97it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.90it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.86it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.84it/s]\n 23%|██▎ | 7/30 [00:01<00:06, 3.83it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.82it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.81it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.81it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.81it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.81it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.80it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.80it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.80it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.80it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.80it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.80it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.79it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.79it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.80it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s]\n 77%|███████▋ | 23/30 [00:06<00:01, 3.80it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.80it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.80it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.80it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.80it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.80it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.7113133, "total_time": 9.040513 }, "output": [ "https://replicate.delivery/xezq/ivbljTJxZ7YyMFAGcIDslCx4nUtHlHNkOSKmfOLFoAfBMXkUA/out-0.webp" ], "started_at": "2025-04-19T15:27:53.208200Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-4k4hl46ogj2fh53gu3cvwnxnsq4inmvvduek5nkoig7e7vkahdta", "get": "https://api.replicate.com/v1/predictions/rkc63xyc5xrma0cp9tmtmj534m", "cancel": "https://api.replicate.com/v1/predictions/rkc63xyc5xrma0cp9tmtmj534m/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 1225 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.73it/s] 7%|▋ | 2/30 [00:00<00:06, 4.11it/s] 10%|█ | 3/30 [00:00<00:06, 3.97it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.90it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.86it/s] 20%|██ | 6/30 [00:01<00:06, 3.84it/s] 23%|██▎ | 7/30 [00:01<00:06, 3.83it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.82it/s] 30%|███ | 9/30 [00:02<00:05, 3.81it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.81it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.81it/s] 40%|████ | 12/30 [00:03<00:04, 3.81it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.80it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.80it/s] 50%|█████ | 15/30 [00:03<00:03, 3.80it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.80it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.80it/s] 60%|██████ | 18/30 [00:04<00:03, 3.80it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.79it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.79it/s] 70%|███████ | 21/30 [00:05<00:02, 3.80it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s] 77%|███████▋ | 23/30 [00:06<00:01, 3.80it/s] 80%|████████ | 24/30 [00:06<00:01, 3.80it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.80it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.80it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.80it/s] 100%|██████████| 30/30 [00:07<00:00, 3.80it/s] 100%|██████████| 30/30 [00:07<00:00, 3.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDbjeht8r5y5rme0cp9tnrbb5rp8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.36
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.36, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.36, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.36, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.36, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.36' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.36, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:29:21.859799Z", "created_at": "2025-04-19T15:29:13.201000Z", "data_removed": false, "error": null, "id": "bjeht8r5y5rme0cp9tnrbb5rp8", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.36, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 61603\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.75it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.15it/s]\n 10%|█ | 3/30 [00:00<00:06, 4.00it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.94it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.90it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.89it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.87it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.86it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.85it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.84it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.84it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.84it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.84it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.84it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.84it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.84it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.85it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.85it/s]\n 77%|███████▋ | 23/30 [00:05<00:01, 3.84it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.84it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.84it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.84it/s]\n 90%|█████████ | 27/30 [00:06<00:00, 3.84it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.84it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.85it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.84it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.86it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.631174496, "total_time": 8.658799 }, "output": [ "https://replicate.delivery/xezq/H44vKIWw4xq1G1bwNwACZSAJYrwMlJBoenqkjf0r0EKRNXkUA/out-0.webp" ], "started_at": "2025-04-19T15:29:13.228624Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-evsrrnak76dux5htnfqlvxbslqbtquzxjmmx5g42y4mvrybm4jkq", "get": "https://api.replicate.com/v1/predictions/bjeht8r5y5rme0cp9tnrbb5rp8", "cancel": "https://api.replicate.com/v1/predictions/bjeht8r5y5rme0cp9tnrbb5rp8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 61603 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.75it/s] 7%|▋ | 2/30 [00:00<00:06, 4.15it/s] 10%|█ | 3/30 [00:00<00:06, 4.00it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.94it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.90it/s] 20%|██ | 6/30 [00:01<00:06, 3.89it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.87it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.86it/s] 30%|███ | 9/30 [00:02<00:05, 3.85it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s] 40%|████ | 12/30 [00:03<00:04, 3.84it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s] 50%|█████ | 15/30 [00:03<00:03, 3.84it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.84it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.84it/s] 60%|██████ | 18/30 [00:04<00:03, 3.84it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.84it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.84it/s] 70%|███████ | 21/30 [00:05<00:02, 3.85it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.85it/s] 77%|███████▋ | 23/30 [00:05<00:01, 3.84it/s] 80%|████████ | 24/30 [00:06<00:01, 3.84it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.84it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.84it/s] 90%|█████████ | 27/30 [00:06<00:00, 3.84it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.84it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.85it/s] 100%|██████████| 30/30 [00:07<00:00, 3.84it/s] 100%|██████████| 30/30 [00:07<00:00, 3.86it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDtrdemztzfxrmc0cp9tyvw7cft0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 1.64
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 1.64, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 1.64, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 1.64, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 1.64, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=1.64' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \\nStyle: crystalline fantasy \\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \\nMood/Atmosphere: tranquil mystery, submerged elegance \\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 1.64, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:49:24.788712Z", "created_at": "2025-04-19T15:49:15.775000Z", "data_removed": false, "error": null, "id": "trdemztzfxrmc0cp9tyvw7cft0", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead \nStyle: crystalline fantasy \nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her \nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin \nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold \nMood/Atmosphere: tranquil mystery, submerged elegance \nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments \nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 1.64, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 29139\nPrompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead\nStyle: crystalline fantasy\nComposition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her\nLighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin\nColor Palette: cool aquamarines, soft lavenders, flickers of rose‑gold\nMood/Atmosphere: tranquil mystery, submerged elegance\nTechnical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments\nAdditional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.74it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.14it/s]\n 10%|█ | 3/30 [00:00<00:06, 4.00it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.90it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.86it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.83it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.83it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.83it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.83it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.82it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.82it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.83it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.83it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.82it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.82it/s]\n 77%|███████▋ | 23/30 [00:05<00:01, 3.81it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.82it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.82it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.81it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.82it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.678241752, "total_time": 9.013712 }, "output": [ "https://replicate.delivery/xezq/QoPjP3WMVd5DLd9LbjpEPgMzmNUDO7b8LbLm9OXkHMMB4FJF/out-0.webp" ], "started_at": "2025-04-19T15:49:16.110470Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-ihofz6gsayv5eujy4nd42kdwosvrcw7aetrz7zapntb6jpmif2tq", "get": "https://api.replicate.com/v1/predictions/trdemztzfxrmc0cp9tyvw7cft0", "cancel": "https://api.replicate.com/v1/predictions/trdemztzfxrmc0cp9tyvw7cft0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 29139 Prompt: Subject: TOKMAHI’s face and neck visible in tight portrait, looking serenely ahead Style: crystalline fantasy Composition: she emerges from a submerged glass cathedral, shards of colored glass floating like stained‑glass fish around her Lighting: refracted light beams breaking into prism rainbows across her features, subtle rim glints on hair and skin Color Palette: cool aquamarines, soft lavenders, flickers of rose‑gold Mood/Atmosphere: tranquil mystery, submerged elegance Technical Details: tack‑sharp focus on her eyes and skin, gentle motion blur on drifting glass fragments Additional Elements: faint ripples of water distortion, bubbles rising into inverted arches overhead --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.74it/s] 7%|▋ | 2/30 [00:00<00:06, 4.14it/s] 10%|█ | 3/30 [00:00<00:06, 4.00it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.90it/s] 20%|██ | 6/30 [00:01<00:06, 3.86it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s] 30%|███ | 9/30 [00:02<00:05, 3.83it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.83it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.83it/s] 40%|████ | 12/30 [00:03<00:04, 3.83it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.82it/s] 50%|█████ | 15/30 [00:03<00:03, 3.82it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.83it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s] 60%|██████ | 18/30 [00:04<00:03, 3.83it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s] 70%|███████ | 21/30 [00:05<00:02, 3.82it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.82it/s] 77%|███████▋ | 23/30 [00:05<00:01, 3.81it/s] 80%|████████ | 24/30 [00:06<00:01, 3.82it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.82it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.81it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.82it/s] 100%|██████████| 30/30 [00:07<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID4heaj556z1rma0cp9v1sdvzafwStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence Style: enchanted surrealism Composition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light Lighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair Color Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white Mood/Atmosphere: mystical, intimate, otherworldly serenity Technical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps Additional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 7.88
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 44
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 7.88, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 7.88, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 44 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 7.88, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 7.88, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=7.88' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=44'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 7.88, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T15:56:19.764125Z", "created_at": "2025-04-19T15:56:07.288000Z", "data_removed": false, "error": null, "id": "4heaj556z1rma0cp9v1sdvzafw", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 7.88, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 44 }, "logs": "Loaded LoRAs in 0.59s\nUsing seed: 62190\nPrompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence\nStyle: enchanted surrealism\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white\nMood/Atmosphere: mystical, intimate, otherworldly serenity\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop\n[!] txt2img mode\n 0%| | 0/44 [00:00<?, ?it/s]\n 2%|▏ | 1/44 [00:00<00:11, 3.71it/s]\n 5%|▍ | 2/44 [00:00<00:10, 4.13it/s]\n 7%|▋ | 3/44 [00:00<00:10, 3.96it/s]\n 9%|▉ | 4/44 [00:01<00:10, 3.90it/s]\n 11%|█▏ | 5/44 [00:01<00:10, 3.87it/s]\n 14%|█▎ | 6/44 [00:01<00:09, 3.85it/s]\n 16%|█▌ | 7/44 [00:01<00:09, 3.82it/s]\n 18%|█▊ | 8/44 [00:02<00:09, 3.82it/s]\n 20%|██ | 9/44 [00:02<00:09, 3.81it/s]\n 23%|██▎ | 10/44 [00:02<00:08, 3.81it/s]\n 25%|██▌ | 11/44 [00:02<00:08, 3.80it/s]\n 27%|██▋ | 12/44 [00:03<00:08, 3.81it/s]\n 30%|██▉ | 13/44 [00:03<00:08, 3.81it/s]\n 32%|███▏ | 14/44 [00:03<00:07, 3.80it/s]\n 34%|███▍ | 15/44 [00:03<00:07, 3.79it/s]\n 36%|███▋ | 16/44 [00:04<00:07, 3.81it/s]\n 39%|███▊ | 17/44 [00:04<00:07, 3.80it/s]\n 41%|████ | 18/44 [00:04<00:06, 3.80it/s]\n 43%|████▎ | 19/44 [00:04<00:06, 3.80it/s]\n 45%|████▌ | 20/44 [00:05<00:06, 3.80it/s]\n 48%|████▊ | 21/44 [00:05<00:06, 3.80it/s]\n 50%|█████ | 22/44 [00:05<00:05, 3.79it/s]\n 52%|█████▏ | 23/44 [00:06<00:05, 3.80it/s]\n 55%|█████▍ | 24/44 [00:06<00:05, 3.80it/s]\n 57%|█████▋ | 25/44 [00:06<00:04, 3.80it/s]\n 59%|█████▉ | 26/44 [00:06<00:04, 3.79it/s]\n 61%|██████▏ | 27/44 [00:07<00:04, 3.80it/s]\n 64%|██████▎ | 28/44 [00:07<00:04, 3.80it/s]\n 66%|██████▌ | 29/44 [00:07<00:03, 3.80it/s]\n 68%|██████▊ | 30/44 [00:07<00:03, 3.80it/s]\n 70%|███████ | 31/44 [00:08<00:03, 3.80it/s]\n 73%|███████▎ | 32/44 [00:08<00:03, 3.80it/s]\n 75%|███████▌ | 33/44 [00:08<00:02, 3.80it/s]\n 77%|███████▋ | 34/44 [00:08<00:02, 3.79it/s]\n 80%|███████▉ | 35/44 [00:09<00:02, 3.80it/s]\n 82%|████████▏ | 36/44 [00:09<00:02, 3.80it/s]\n 84%|████████▍ | 37/44 [00:09<00:01, 3.80it/s]\n 86%|████████▋ | 38/44 [00:09<00:01, 3.79it/s]\n 89%|████████▊ | 39/44 [00:10<00:01, 3.80it/s]\n 91%|█████████ | 40/44 [00:10<00:01, 3.80it/s]\n 93%|█████████▎| 41/44 [00:10<00:00, 3.80it/s]\n 95%|█████████▌| 42/44 [00:11<00:00, 3.80it/s]\n 98%|█████████▊| 43/44 [00:11<00:00, 3.80it/s]\n100%|██████████| 44/44 [00:11<00:00, 3.80it/s]\n100%|██████████| 44/44 [00:11<00:00, 3.81it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 12.423833537, "total_time": 12.476125 }, "output": [ "https://replicate.delivery/xezq/BSoIBXH3VDbUJFIwnaQh0YFVrT5s4rRujnWhCeWMoHvRzLSKA/out-0.webp" ], "started_at": "2025-04-19T15:56:07.340292Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-7xsvwgrkv6tp2iirvfjvrqfgqs2dzir3uxvx3xftyd6rmepiftvq", "get": "https://api.replicate.com/v1/predictions/4heaj556z1rma0cp9v1sdvzafw", "cancel": "https://api.replicate.com/v1/predictions/4heaj556z1rma0cp9v1sdvzafw/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.59s Using seed: 62190 Prompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence Style: enchanted surrealism Composition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light Lighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair Color Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white Mood/Atmosphere: mystical, intimate, otherworldly serenity Technical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps Additional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop [!] txt2img mode 0%| | 0/44 [00:00<?, ?it/s] 2%|▏ | 1/44 [00:00<00:11, 3.71it/s] 5%|▍ | 2/44 [00:00<00:10, 4.13it/s] 7%|▋ | 3/44 [00:00<00:10, 3.96it/s] 9%|▉ | 4/44 [00:01<00:10, 3.90it/s] 11%|█▏ | 5/44 [00:01<00:10, 3.87it/s] 14%|█▎ | 6/44 [00:01<00:09, 3.85it/s] 16%|█▌ | 7/44 [00:01<00:09, 3.82it/s] 18%|█▊ | 8/44 [00:02<00:09, 3.82it/s] 20%|██ | 9/44 [00:02<00:09, 3.81it/s] 23%|██▎ | 10/44 [00:02<00:08, 3.81it/s] 25%|██▌ | 11/44 [00:02<00:08, 3.80it/s] 27%|██▋ | 12/44 [00:03<00:08, 3.81it/s] 30%|██▉ | 13/44 [00:03<00:08, 3.81it/s] 32%|███▏ | 14/44 [00:03<00:07, 3.80it/s] 34%|███▍ | 15/44 [00:03<00:07, 3.79it/s] 36%|███▋ | 16/44 [00:04<00:07, 3.81it/s] 39%|███▊ | 17/44 [00:04<00:07, 3.80it/s] 41%|████ | 18/44 [00:04<00:06, 3.80it/s] 43%|████▎ | 19/44 [00:04<00:06, 3.80it/s] 45%|████▌ | 20/44 [00:05<00:06, 3.80it/s] 48%|████▊ | 21/44 [00:05<00:06, 3.80it/s] 50%|█████ | 22/44 [00:05<00:05, 3.79it/s] 52%|█████▏ | 23/44 [00:06<00:05, 3.80it/s] 55%|█████▍ | 24/44 [00:06<00:05, 3.80it/s] 57%|█████▋ | 25/44 [00:06<00:04, 3.80it/s] 59%|█████▉ | 26/44 [00:06<00:04, 3.79it/s] 61%|██████▏ | 27/44 [00:07<00:04, 3.80it/s] 64%|██████▎ | 28/44 [00:07<00:04, 3.80it/s] 66%|██████▌ | 29/44 [00:07<00:03, 3.80it/s] 68%|██████▊ | 30/44 [00:07<00:03, 3.80it/s] 70%|███████ | 31/44 [00:08<00:03, 3.80it/s] 73%|███████▎ | 32/44 [00:08<00:03, 3.80it/s] 75%|███████▌ | 33/44 [00:08<00:02, 3.80it/s] 77%|███████▋ | 34/44 [00:08<00:02, 3.79it/s] 80%|███████▉ | 35/44 [00:09<00:02, 3.80it/s] 82%|████████▏ | 36/44 [00:09<00:02, 3.80it/s] 84%|████████▍ | 37/44 [00:09<00:01, 3.80it/s] 86%|████████▋ | 38/44 [00:09<00:01, 3.79it/s] 89%|████████▊ | 39/44 [00:10<00:01, 3.80it/s] 91%|█████████ | 40/44 [00:10<00:01, 3.80it/s] 93%|█████████▎| 41/44 [00:10<00:00, 3.80it/s] 95%|█████████▌| 42/44 [00:11<00:00, 3.80it/s] 98%|█████████▊| 43/44 [00:11<00:00, 3.80it/s] 100%|██████████| 44/44 [00:11<00:00, 3.80it/s] 100%|██████████| 44/44 [00:11<00:00, 3.81it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDvemga3egxnrme0cp9wxbcpjsfwStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85
- go_fast
- lora_scale
- 1.01
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", go_fast: false, lora_scale: 1.01, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": False, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt=" TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.01' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:06:26.992962Z", "created_at": "2025-04-19T18:06:16.813000Z", "data_removed": false, "error": null, "id": "vemga3egxnrme0cp9wxbcpjsfw", "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "free=24284863893504\nDownloading weights\n2025-04-19T18:06:16Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp5m5euhrr/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T18:06:16Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T18:06:18Z | INFO | [ Complete ] dest=/tmp/tmp5m5euhrr/weights size=\"172 MB\" total_elapsed=1.405s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 1.42s\nLoaded LoRAs in 1.98s\nUsing seed: 54992\nPrompt: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction\nStyle: avant‑garde deconstructive abstraction\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents\nMood/Atmosphere: dissonant elegance, fractured surrealism\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.72it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.14it/s]\n 10%|█ | 3/30 [00:00<00:06, 3.99it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.87it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.86it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.84it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.85it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.84it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.83it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.82it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.84it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.84it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.84it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.83it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.83it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s]\n 77%|███████▋ | 23/30 [00:05<00:01, 3.83it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.83it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.83it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.83it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.85it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.061602688, "total_time": 10.179962 }, "output": [ "https://replicate.delivery/xezq/r9MBbIm4907GOBS9evao2MEoWepiafbyrLtwCKKbV6rFBzIpA/out-0.webp" ], "started_at": "2025-04-19T18:06:16.931359Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-5g6ge365cy2fldnjjpzmb4rattgcjdegkxlf5pdjktx4xoqii7ca", "get": "https://api.replicate.com/v1/predictions/vemga3egxnrme0cp9wxbcpjsfw", "cancel": "https://api.replicate.com/v1/predictions/vemga3egxnrme0cp9wxbcpjsfw/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=24284863893504 Downloading weights 2025-04-19T18:06:16Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp5m5euhrr/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T18:06:16Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T18:06:18Z | INFO | [ Complete ] dest=/tmp/tmp5m5euhrr/weights size="172 MB" total_elapsed=1.405s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 1.42s Loaded LoRAs in 1.98s Using seed: 54992 Prompt: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.72it/s] 7%|▋ | 2/30 [00:00<00:06, 4.14it/s] 10%|█ | 3/30 [00:00<00:06, 3.99it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s] 20%|██ | 6/30 [00:01<00:06, 3.87it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.86it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.84it/s] 30%|███ | 9/30 [00:02<00:05, 3.85it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s] 40%|████ | 12/30 [00:03<00:04, 3.84it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s] 50%|█████ | 15/30 [00:03<00:03, 3.83it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.82it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.84it/s] 60%|██████ | 18/30 [00:04<00:03, 3.84it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.84it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.83it/s] 70%|███████ | 21/30 [00:05<00:02, 3.83it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s] 77%|███████▋ | 23/30 [00:05<00:01, 3.83it/s] 80%|████████ | 24/30 [00:06<00:01, 3.83it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.83it/s] 100%|██████████| 30/30 [00:07<00:00, 3.83it/s] 100%|██████████| 30/30 [00:07<00:00, 3.85it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDkqyn2bnqgxrme0cp9x0tkg2nawStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches Style: avant‑garde deconstructive abstraction Composition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs Lighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares Color Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber Mood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC Technical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges Additional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags --ar 1:1 --v flux1 --q 2 --style raw --chaos 85
- go_fast
- lora_scale
- 1.019
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.18
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 34
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", go_fast: false, lora_scale: 1.019, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.18, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 34 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": False, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.019' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.18' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=34'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:13:58.261305Z", "created_at": "2025-04-19T18:13:49.063000Z", "data_removed": false, "error": null, "id": "kqyn2bnqgxrme0cp9x0tkg2naw", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 34 }, "logs": "Weights already loaded\nLoaded LoRAs in 0.03s\nUsing seed: 2755\nPrompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches\nStyle: avant‑garde deconstructive abstraction\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n[!] txt2img mode\n 0%| | 0/34 [00:00<?, ?it/s]\n 3%|▎ | 1/34 [00:00<00:08, 3.71it/s]\n 6%|▌ | 2/34 [00:00<00:07, 4.13it/s]\n 9%|▉ | 3/34 [00:00<00:07, 3.98it/s]\n 12%|█▏ | 4/34 [00:01<00:07, 3.90it/s]\n 15%|█▍ | 5/34 [00:01<00:07, 3.87it/s]\n 18%|█▊ | 6/34 [00:01<00:07, 3.86it/s]\n 21%|██ | 7/34 [00:01<00:07, 3.84it/s]\n 24%|██▎ | 8/34 [00:02<00:06, 3.83it/s]\n 26%|██▋ | 9/34 [00:02<00:06, 3.83it/s]\n 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s]\n 32%|███▏ | 11/34 [00:02<00:06, 3.82it/s]\n 35%|███▌ | 12/34 [00:03<00:05, 3.82it/s]\n 38%|███▊ | 13/34 [00:03<00:05, 3.82it/s]\n 41%|████ | 14/34 [00:03<00:05, 3.83it/s]\n 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 16/34 [00:04<00:04, 3.82it/s]\n 50%|█████ | 17/34 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 18/34 [00:04<00:04, 3.83it/s]\n 56%|█████▌ | 19/34 [00:04<00:03, 3.81it/s]\n 59%|█████▉ | 20/34 [00:05<00:03, 3.82it/s]\n 62%|██████▏ | 21/34 [00:05<00:03, 3.83it/s]\n 65%|██████▍ | 22/34 [00:05<00:03, 3.83it/s]\n 68%|██████▊ | 23/34 [00:05<00:02, 3.82it/s]\n 71%|███████ | 24/34 [00:06<00:02, 3.81it/s]\n 74%|███████▎ | 25/34 [00:06<00:02, 3.82it/s]\n 76%|███████▋ | 26/34 [00:06<00:02, 3.81it/s]\n 79%|███████▉ | 27/34 [00:07<00:01, 3.81it/s]\n 82%|████████▏ | 28/34 [00:07<00:01, 3.82it/s]\n 85%|████████▌ | 29/34 [00:07<00:01, 3.83it/s]\n 88%|████████▊ | 30/34 [00:07<00:01, 3.82it/s]\n 91%|█████████ | 31/34 [00:08<00:00, 3.82it/s]\n 94%|█████████▍| 32/34 [00:08<00:00, 3.83it/s]\n 97%|█████████▋| 33/34 [00:08<00:00, 3.82it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.82it/s]\n100%|██████████| 34/34 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.190657339, "total_time": 9.198305 }, "output": [ "https://replicate.delivery/xezq/wGISM2b2yrJjKRlFKBAbOMIWsmCrNwelGPWY4VN8XWOzzMSKA/out-0.webp" ], "started_at": "2025-04-19T18:13:49.070648Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-3hgy7vlw3wr255jlvqb7ht7uqmsofhag5v65lkariqf7kkkkzywq", "get": "https://api.replicate.com/v1/predictions/kqyn2bnqgxrme0cp9x0tkg2naw", "cancel": "https://api.replicate.com/v1/predictions/kqyn2bnqgxrme0cp9x0tkg2naw/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inWeights already loaded Loaded LoRAs in 0.03s Using seed: 2755 Prompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches Style: avant‑garde deconstructive abstraction Composition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs Lighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares Color Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber Mood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC Technical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges Additional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags --ar 1:1 --v flux1 --q 2 --style raw --chaos 85 [!] txt2img mode 0%| | 0/34 [00:00<?, ?it/s] 3%|▎ | 1/34 [00:00<00:08, 3.71it/s] 6%|▌ | 2/34 [00:00<00:07, 4.13it/s] 9%|▉ | 3/34 [00:00<00:07, 3.98it/s] 12%|█▏ | 4/34 [00:01<00:07, 3.90it/s] 15%|█▍ | 5/34 [00:01<00:07, 3.87it/s] 18%|█▊ | 6/34 [00:01<00:07, 3.86it/s] 21%|██ | 7/34 [00:01<00:07, 3.84it/s] 24%|██▎ | 8/34 [00:02<00:06, 3.83it/s] 26%|██▋ | 9/34 [00:02<00:06, 3.83it/s] 29%|██▉ | 10/34 [00:02<00:06, 3.83it/s] 32%|███▏ | 11/34 [00:02<00:06, 3.82it/s] 35%|███▌ | 12/34 [00:03<00:05, 3.82it/s] 38%|███▊ | 13/34 [00:03<00:05, 3.82it/s] 41%|████ | 14/34 [00:03<00:05, 3.83it/s] 44%|████▍ | 15/34 [00:03<00:04, 3.82it/s] 47%|████▋ | 16/34 [00:04<00:04, 3.82it/s] 50%|█████ | 17/34 [00:04<00:04, 3.82it/s] 53%|█████▎ | 18/34 [00:04<00:04, 3.83it/s] 56%|█████▌ | 19/34 [00:04<00:03, 3.81it/s] 59%|█████▉ | 20/34 [00:05<00:03, 3.82it/s] 62%|██████▏ | 21/34 [00:05<00:03, 3.83it/s] 65%|██████▍ | 22/34 [00:05<00:03, 3.83it/s] 68%|██████▊ | 23/34 [00:05<00:02, 3.82it/s] 71%|███████ | 24/34 [00:06<00:02, 3.81it/s] 74%|███████▎ | 25/34 [00:06<00:02, 3.82it/s] 76%|███████▋ | 26/34 [00:06<00:02, 3.81it/s] 79%|███████▉ | 27/34 [00:07<00:01, 3.81it/s] 82%|████████▏ | 28/34 [00:07<00:01, 3.82it/s] 85%|████████▌ | 29/34 [00:07<00:01, 3.83it/s] 88%|████████▊ | 30/34 [00:07<00:01, 3.82it/s] 91%|█████████ | 31/34 [00:08<00:00, 3.82it/s] 94%|█████████▍| 32/34 [00:08<00:00, 3.83it/s] 97%|█████████▋| 33/34 [00:08<00:00, 3.82it/s] 100%|██████████| 34/34 [00:08<00:00, 3.82it/s] 100%|██████████| 34/34 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID3kp6z6vdfxrmc0cp9xp9jx5dg0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- TOKMAHI captured mid‑motion, hair and fabric trailing dynamically Style: cinematic motion Composition: medium portrait from waist up, slight pan‑blur horizontally Lighting: strong key light from one side with fill to retain facial detail Color Palette: muted cinematic tones—teal shadows, warm highlights Mood/Atmosphere: energetic, in‑flight momentum Technical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur Additional Elements: trailing blur on hair and coat edges, light streaks in background --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \nStyle: cinematic motion \nComposition: medium portrait from waist up, slight pan‑blur horizontally \nLighting: strong key light from one side with fill to retain facial detail \nColor Palette: muted cinematic tones—teal shadows, warm highlights \nMood/Atmosphere: energetic, in‑flight momentum \nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \nStyle: cinematic motion \nComposition: medium portrait from waist up, slight pan‑blur horizontally \nLighting: strong key light from one side with fill to retain facial detail \nColor Palette: muted cinematic tones—teal shadows, warm highlights \nMood/Atmosphere: energetic, in‑flight momentum \nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \nStyle: cinematic motion \nComposition: medium portrait from waist up, slight pan‑blur horizontally \nLighting: strong key light from one side with fill to retain facial detail \nColor Palette: muted cinematic tones—teal shadows, warm highlights \nMood/Atmosphere: energetic, in‑flight momentum \nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \\nStyle: cinematic motion \\nComposition: medium portrait from waist up, slight pan‑blur horizontally \\nLighting: strong key light from one side with fill to retain facial detail \\nColor Palette: muted cinematic tones—teal shadows, warm highlights \\nMood/Atmosphere: energetic, in‑flight momentum \\nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \\nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt=" TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \\nStyle: cinematic motion \\nComposition: medium portrait from waist up, slight pan‑blur horizontally \\nLighting: strong key light from one side with fill to retain facial detail \\nColor Palette: muted cinematic tones—teal shadows, warm highlights \\nMood/Atmosphere: energetic, in‑flight momentum \\nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \\nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \\nStyle: cinematic motion \\nComposition: medium portrait from waist up, slight pan‑blur horizontally \\nLighting: strong key light from one side with fill to retain facial detail \\nColor Palette: muted cinematic tones—teal shadows, warm highlights \\nMood/Atmosphere: energetic, in‑flight momentum \\nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \\nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:00:38.725209Z", "created_at": "2025-04-19T19:00:28.159000Z", "data_removed": false, "error": null, "id": "3kp6z6vdfxrmc0cp9xp9jx5dg0", "input": { "model": "dev", "prompt": " TOKMAHI captured mid‑motion, hair and fabric trailing dynamically \nStyle: cinematic motion \nComposition: medium portrait from waist up, slight pan‑blur horizontally \nLighting: strong key light from one side with fill to retain facial detail \nColor Palette: muted cinematic tones—teal shadows, warm highlights \nMood/Atmosphere: energetic, in‑flight momentum \nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur \nAdditional Elements: trailing blur on hair and coat edges, light streaks in background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "free=28908610035712\nDownloading weights\n2025-04-19T19:00:28Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp0zf5ckbs/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T19:00:28Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T19:00:30Z | INFO | [ Complete ] dest=/tmp/tmp0zf5ckbs/weights size=\"172 MB\" total_elapsed=1.492s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 1.51s\nLoaded LoRAs in 2.08s\nUsing seed: 24775\nPrompt: TOKMAHI captured mid‑motion, hair and fabric trailing dynamically\nStyle: cinematic motion\nComposition: medium portrait from waist up, slight pan‑blur horizontally\nLighting: strong key light from one side with fill to retain facial detail\nColor Palette: muted cinematic tones—teal shadows, warm highlights\nMood/Atmosphere: energetic, in‑flight momentum\nTechnical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur\nAdditional Elements: trailing blur on hair and coat edges, light streaks in background\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.74it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.13it/s]\n 10%|█ | 3/30 [00:00<00:06, 3.99it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.86it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.84it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.84it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.83it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.83it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.83it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.82it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.82it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.83it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.82it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.82it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.82it/s]\n 77%|███████▋ | 23/30 [00:05<00:01, 3.82it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.82it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.83it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.82it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.84it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.165233749, "total_time": 10.566209 }, "output": [ "https://replicate.delivery/xezq/3MQHyTfWQsxPDSvagaC1A8YiDXnsryKHLNfL4sJysm4WTakUA/out-0.webp" ], "started_at": "2025-04-19T19:00:28.559975Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-h66l57pqvfa3cw2zlhdrv3mimkjrndn5qzfvc7akxejzo65bho3a", "get": "https://api.replicate.com/v1/predictions/3kp6z6vdfxrmc0cp9xp9jx5dg0", "cancel": "https://api.replicate.com/v1/predictions/3kp6z6vdfxrmc0cp9xp9jx5dg0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=28908610035712 Downloading weights 2025-04-19T19:00:28Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp0zf5ckbs/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T19:00:28Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T19:00:30Z | INFO | [ Complete ] dest=/tmp/tmp0zf5ckbs/weights size="172 MB" total_elapsed=1.492s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 1.51s Loaded LoRAs in 2.08s Using seed: 24775 Prompt: TOKMAHI captured mid‑motion, hair and fabric trailing dynamically Style: cinematic motion Composition: medium portrait from waist up, slight pan‑blur horizontally Lighting: strong key light from one side with fill to retain facial detail Color Palette: muted cinematic tones—teal shadows, warm highlights Mood/Atmosphere: energetic, in‑flight momentum Technical Details: 35 mm lens simulation, f/2.8, simulated 1/20 s shutter for motion blur Additional Elements: trailing blur on hair and coat edges, light streaks in background --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 50 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.74it/s] 7%|▋ | 2/30 [00:00<00:06, 4.13it/s] 10%|█ | 3/30 [00:00<00:06, 3.99it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s] 20%|██ | 6/30 [00:01<00:06, 3.86it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.84it/s] 30%|███ | 9/30 [00:02<00:05, 3.84it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.83it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.83it/s] 40%|████ | 12/30 [00:03<00:04, 3.83it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.82it/s] 50%|█████ | 15/30 [00:03<00:03, 3.82it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.83it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s] 60%|██████ | 18/30 [00:04<00:03, 3.82it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s] 70%|███████ | 21/30 [00:05<00:02, 3.82it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.82it/s] 77%|███████▋ | 23/30 [00:05<00:01, 3.82it/s] 80%|████████ | 24/30 [00:06<00:01, 3.82it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.83it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.82it/s] 100%|██████████| 30/30 [00:07<00:00, 3.84it/s] 100%|██████████| 30/30 [00:07<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDvb5r824x3drmc0cp9xy88wet1gStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged Style: high‑key beauty elegance Composition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background Lighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows Color Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents Mood/Atmosphere: fresh, airy, flawless luminosity Technical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh Additional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "model": "dev", "prompt": "\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \nStyle: high‑key beauty elegance \nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \nStyle: high‑key beauty elegance \nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \nStyle: high‑key beauty elegance \nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \\nStyle: high‑key beauty elegance \\nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \\nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \\nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \\nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \\nStyle: high‑key beauty elegance \\nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \\nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \\nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \\nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \\nStyle: high‑key beauty elegance \\nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \\nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \\nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \\nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:18:20.127482Z", "created_at": "2025-04-19T19:18:08.923000Z", "data_removed": false, "error": null, "id": "vb5r824x3drmc0cp9xy88wet1g", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged \nStyle: high‑key beauty elegance \nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background \nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows \nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh \nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 52708\nPrompt:\nSubject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged\nStyle: high‑key beauty elegance\nComposition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background\nLighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows\nColor Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents\nMood/Atmosphere: fresh, airy, flawless luminosity\nTechnical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh\nAdditional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20\n[!] txt2img mode\n 0%| | 0/38 [00:00<?, ?it/s]\n 3%|▎ | 1/38 [00:00<00:09, 3.72it/s]\n 5%|▌ | 2/38 [00:00<00:08, 4.13it/s]\n 8%|▊ | 3/38 [00:00<00:08, 3.97it/s]\n 11%|█ | 4/38 [00:01<00:08, 3.90it/s]\n 13%|█▎ | 5/38 [00:01<00:08, 3.87it/s]\n 16%|█▌ | 6/38 [00:01<00:08, 3.85it/s]\n 18%|█▊ | 7/38 [00:01<00:08, 3.84it/s]\n 21%|██ | 8/38 [00:02<00:07, 3.82it/s]\n 24%|██▎ | 9/38 [00:02<00:07, 3.82it/s]\n 26%|██▋ | 10/38 [00:02<00:07, 3.82it/s]\n 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s]\n 32%|███▏ | 12/38 [00:03<00:06, 3.81it/s]\n 34%|███▍ | 13/38 [00:03<00:06, 3.81it/s]\n 37%|███▋ | 14/38 [00:03<00:06, 3.81it/s]\n 39%|███▉ | 15/38 [00:03<00:06, 3.81it/s]\n 42%|████▏ | 16/38 [00:04<00:05, 3.80it/s]\n 45%|████▍ | 17/38 [00:04<00:05, 3.80it/s]\n 47%|████▋ | 18/38 [00:04<00:05, 3.80it/s]\n 50%|█████ | 19/38 [00:04<00:04, 3.81it/s]\n 53%|█████▎ | 20/38 [00:05<00:04, 3.79it/s]\n 55%|█████▌ | 21/38 [00:05<00:04, 3.80it/s]\n 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s]\n 61%|██████ | 23/38 [00:06<00:03, 3.82it/s]\n 63%|██████▎ | 24/38 [00:06<00:03, 3.81it/s]\n 66%|██████▌ | 25/38 [00:06<00:03, 3.81it/s]\n 68%|██████▊ | 26/38 [00:06<00:03, 3.81it/s]\n 71%|███████ | 27/38 [00:07<00:02, 3.81it/s]\n 74%|███████▎ | 28/38 [00:07<00:02, 3.81it/s]\n 76%|███████▋ | 29/38 [00:07<00:02, 3.81it/s]\n 79%|███████▉ | 30/38 [00:07<00:02, 3.81it/s]\n 82%|████████▏ | 31/38 [00:08<00:01, 3.80it/s]\n 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s]\n 87%|████████▋ | 33/38 [00:08<00:01, 3.81it/s]\n 89%|████████▉ | 34/38 [00:08<00:01, 3.81it/s]\n 92%|█████████▏| 35/38 [00:09<00:00, 3.80it/s]\n 95%|█████████▍| 36/38 [00:09<00:00, 3.80it/s]\n 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.81it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.784931203, "total_time": 11.204482 }, "output": [ "https://replicate.delivery/xezq/j6Cp0xpyB14CEdfHnjGugeIerOf2VJfXJlQMCnlB5P4kfoGJF/out-0.webp" ], "started_at": "2025-04-19T19:18:09.342551Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-hh5h52uv3tohy7b2pqcqxjnvhwejvu6db3ygzc7ndcpstb2n4pka", "get": "https://api.replicate.com/v1/predictions/vb5r824x3drmc0cp9xy88wet1g", "cancel": "https://api.replicate.com/v1/predictions/vb5r824x3drmc0cp9xy88wet1g/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 52708 Prompt: Subject: TOKMAHI in a mid‑length beauty portrait, shoulders relaxed, head tilted slightly, eyes soft and engaged Style: high‑key beauty elegance Composition: tight crop (waist up), centered—100 mm macro lens simulation, f/2.0, pure white background Lighting: broad soft‑box front‑lighting with under‑fill, creating an ethereal glow and no shadows Color Palette: whites and pastels—blush pink lips, ivory dress, dove‑gray accents Mood/Atmosphere: fresh, airy, flawless luminosity Technical Details: high key exposure, minimal retouching, fine skin texture visible, dreamy bokeh Additional Elements: barely perceptible lens flare, light dusting of iridescent powder on shoulders --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 [!] txt2img mode 0%| | 0/38 [00:00<?, ?it/s] 3%|▎ | 1/38 [00:00<00:09, 3.72it/s] 5%|▌ | 2/38 [00:00<00:08, 4.13it/s] 8%|▊ | 3/38 [00:00<00:08, 3.97it/s] 11%|█ | 4/38 [00:01<00:08, 3.90it/s] 13%|█▎ | 5/38 [00:01<00:08, 3.87it/s] 16%|█▌ | 6/38 [00:01<00:08, 3.85it/s] 18%|█▊ | 7/38 [00:01<00:08, 3.84it/s] 21%|██ | 8/38 [00:02<00:07, 3.82it/s] 24%|██▎ | 9/38 [00:02<00:07, 3.82it/s] 26%|██▋ | 10/38 [00:02<00:07, 3.82it/s] 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s] 32%|███▏ | 12/38 [00:03<00:06, 3.81it/s] 34%|███▍ | 13/38 [00:03<00:06, 3.81it/s] 37%|███▋ | 14/38 [00:03<00:06, 3.81it/s] 39%|███▉ | 15/38 [00:03<00:06, 3.81it/s] 42%|████▏ | 16/38 [00:04<00:05, 3.80it/s] 45%|████▍ | 17/38 [00:04<00:05, 3.80it/s] 47%|████▋ | 18/38 [00:04<00:05, 3.80it/s] 50%|█████ | 19/38 [00:04<00:04, 3.81it/s] 53%|█████▎ | 20/38 [00:05<00:04, 3.79it/s] 55%|█████▌ | 21/38 [00:05<00:04, 3.80it/s] 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s] 61%|██████ | 23/38 [00:06<00:03, 3.82it/s] 63%|██████▎ | 24/38 [00:06<00:03, 3.81it/s] 66%|██████▌ | 25/38 [00:06<00:03, 3.81it/s] 68%|██████▊ | 26/38 [00:06<00:03, 3.81it/s] 71%|███████ | 27/38 [00:07<00:02, 3.81it/s] 74%|███████▎ | 28/38 [00:07<00:02, 3.81it/s] 76%|███████▋ | 29/38 [00:07<00:02, 3.81it/s] 79%|███████▉ | 30/38 [00:07<00:02, 3.81it/s] 82%|████████▏ | 31/38 [00:08<00:01, 3.80it/s] 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s] 87%|████████▋ | 33/38 [00:08<00:01, 3.81it/s] 89%|████████▉ | 34/38 [00:08<00:01, 3.81it/s] 92%|█████████▏| 35/38 [00:09<00:00, 3.80it/s] 95%|█████████▍| 36/38 [00:09<00:00, 3.80it/s] 97%|█████████▋| 37/38 [00:09<00:00, 3.81it/s] 100%|██████████| 38/38 [00:09<00:00, 3.81it/s] 100%|██████████| 38/38 [00:09<00:00, 3.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDzmv10jkw69rmc0cp9xsajge3s8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and shoulders in stark monochrome Style: film noir Composition: tight crop, slight head turn away from light Lighting: strong side light with deep shadows, “Venetian blind” pattern optional Color Palette: pure black & white, no grays Mood/Atmosphere: brooding, suspenseful intrigue Technical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain Additional Elements: subtle smoke curl, reflective glint in her eye --ar 1:1 --v flux1 --q 2 --style raw --chaos 25
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:07:13.902149Z", "created_at": "2025-04-19T19:07:05.138000Z", "data_removed": false, "error": null, "id": "zmv10jkw69rmc0cp9xsajge3s8", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 26655\nPrompt:\nSubject: TOKMAHI’s face and shoulders in stark monochrome\nStyle: film noir\nComposition: tight crop, slight head turn away from light\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional\nColor Palette: pure black & white, no grays\nMood/Atmosphere: brooding, suspenseful intrigue\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain\nAdditional Elements: subtle smoke curl, reflective glint in her eye\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.74it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.14it/s]\n 10%|█ | 3/30 [00:00<00:06, 4.00it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.91it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.86it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.81it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.82it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.82it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.81it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.81it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.81it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.81it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.81it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.81it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.82it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.81it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.81it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.81it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s]\n 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.80it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.683775321, "total_time": 8.764149 }, "output": [ "https://replicate.delivery/xezq/mf4Wl2W6330tCirAziy1J332uaKGfukb15h8NJS34gYhZakUA/out-0.webp" ], "started_at": "2025-04-19T19:07:05.218374Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-7hzldo5mobtamli6wxshakqdiq673m3yuim7a5hialfbrrssydgq", "get": "https://api.replicate.com/v1/predictions/zmv10jkw69rmc0cp9xsajge3s8", "cancel": "https://api.replicate.com/v1/predictions/zmv10jkw69rmc0cp9xsajge3s8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 26655 Prompt: Subject: TOKMAHI’s face and shoulders in stark monochrome Style: film noir Composition: tight crop, slight head turn away from light Lighting: strong side light with deep shadows, “Venetian blind” pattern optional Color Palette: pure black & white, no grays Mood/Atmosphere: brooding, suspenseful intrigue Technical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain Additional Elements: subtle smoke curl, reflective glint in her eye --ar 1:1 --v flux1 --q 2 --style raw --chaos 25 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.74it/s] 7%|▋ | 2/30 [00:00<00:06, 4.14it/s] 10%|█ | 3/30 [00:00<00:06, 4.00it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.91it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s] 20%|██ | 6/30 [00:01<00:06, 3.86it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s] 30%|███ | 9/30 [00:02<00:05, 3.81it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.82it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.82it/s] 40%|████ | 12/30 [00:03<00:04, 3.81it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.81it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.81it/s] 50%|█████ | 15/30 [00:03<00:03, 3.81it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.81it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.81it/s] 60%|██████ | 18/30 [00:04<00:03, 3.82it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.81it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.81it/s] 70%|███████ | 21/30 [00:05<00:02, 3.81it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s] 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s] 80%|████████ | 24/30 [00:06<00:01, 3.80it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID9j3cbgkdg5rme0cp9wxv054sqgStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85
- go_fast
- lora_scale
- 1.01
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", go_fast: false, lora_scale: 1.01, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": False, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt=" TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.01' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:07:06.429045Z", "created_at": "2025-04-19T18:06:56.897000Z", "data_removed": false, "error": null, "id": "9j3cbgkdg5rme0cp9wxv054sqg", "input": { "model": "dev", "prompt": " TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.01, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "free=26269566660608\nDownloading weights\n2025-04-19T18:06:57Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp34lzxa05/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T18:06:57Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T18:06:57Z | INFO | [ Complete ] dest=/tmp/tmp34lzxa05/weights size=\"172 MB\" total_elapsed=0.718s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 0.74s\nLoaded LoRAs in 1.32s\nUsing seed: 52740\nPrompt: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction\nStyle: avant‑garde deconstructive abstraction\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents\nMood/Atmosphere: dissonant elegance, fractured surrealism\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.75it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.15it/s]\n 10%|█ | 3/30 [00:00<00:06, 3.98it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.85it/s]\n 23%|██▎ | 7/30 [00:01<00:06, 3.83it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.81it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.81it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.81it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.80it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.80it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.80it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.80it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.80it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.80it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.79it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.80it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.80it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.81it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.81it/s]\n 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.81it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.82it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.81it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.456945035, "total_time": 9.532045 }, "output": [ "https://replicate.delivery/xezq/3ahgRdjFhf17HSMyNNvneOXwtqIjOUiVUi4hdPmIYg8KhZkUA/out-0.webp" ], "started_at": "2025-04-19T18:06:56.972100Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-yudjnkjf5t7cq6wfhdmunv6d2dqfc5hlswzk6z7dwli4amb4iz3q", "get": "https://api.replicate.com/v1/predictions/9j3cbgkdg5rme0cp9wxv054sqg", "cancel": "https://api.replicate.com/v1/predictions/9j3cbgkdg5rme0cp9wxv054sqg/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=26269566660608 Downloading weights 2025-04-19T18:06:57Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp34lzxa05/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T18:06:57Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T18:06:57Z | INFO | [ Complete ] dest=/tmp/tmp34lzxa05/weights size="172 MB" total_elapsed=0.718s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 0.74s Loaded LoRAs in 1.32s Using seed: 52740 Prompt: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.75it/s] 7%|▋ | 2/30 [00:00<00:06, 4.15it/s] 10%|█ | 3/30 [00:00<00:06, 3.98it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s] 20%|██ | 6/30 [00:01<00:06, 3.85it/s] 23%|██▎ | 7/30 [00:01<00:06, 3.83it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s] 30%|███ | 9/30 [00:02<00:05, 3.81it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.81it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.81it/s] 40%|████ | 12/30 [00:03<00:04, 3.80it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.80it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.80it/s] 50%|█████ | 15/30 [00:03<00:03, 3.80it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.80it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.80it/s] 60%|██████ | 18/30 [00:04<00:03, 3.79it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.80it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.80it/s] 70%|███████ | 21/30 [00:05<00:02, 3.81it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.81it/s] 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s] 80%|████████ | 24/30 [00:06<00:01, 3.81it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.82it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.81it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDwqqv8j8vtnrme0cp9x1r25ew4mStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches Style: avant‑garde deconstructive abstraction Composition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs Lighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares Color Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber Mood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC Technical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges Additional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags --ar 1:1 --v flux1 --q 2 --style raw --chaos 85
- go_fast
- lora_scale
- 1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.18
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 40
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 40 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", go_fast: false, lora_scale: 1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.18, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 40 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": False, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 40 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 40 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.18' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=40'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 40 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:15:31.000212Z", "created_at": "2025-04-19T18:15:20.277000Z", "data_removed": false, "error": null, "id": "wqqv8j8vtnrme0cp9x1r25ew4m", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs \nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares \nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber \nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC \nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges \nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.18, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 40 }, "logs": "Weights already loaded\nLoaded LoRAs in 0.03s\nUsing seed: 39695\nPrompt: Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches\nStyle: avant‑garde deconstructive abstraction\nComposition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs\nLighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares\nColor Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber\nMood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC\nTechnical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges\nAdditional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n[!] txt2img mode\n 0%| | 0/40 [00:00<?, ?it/s]\n 2%|▎ | 1/40 [00:00<00:10, 3.82it/s]\n 5%|▌ | 2/40 [00:00<00:08, 4.33it/s]\n 8%|▊ | 3/40 [00:00<00:09, 4.09it/s]\n 10%|█ | 4/40 [00:00<00:09, 3.99it/s]\n 12%|█▎ | 5/40 [00:01<00:08, 3.93it/s]\n 15%|█▌ | 6/40 [00:01<00:08, 3.90it/s]\n 18%|█▊ | 7/40 [00:01<00:08, 3.88it/s]\n 20%|██ | 8/40 [00:02<00:08, 3.86it/s]\n 22%|██▎ | 9/40 [00:02<00:08, 3.86it/s]\n 25%|██▌ | 10/40 [00:02<00:07, 3.85it/s]\n 28%|██▊ | 11/40 [00:02<00:07, 3.85it/s]\n 30%|███ | 12/40 [00:03<00:07, 3.84it/s]\n 32%|███▎ | 13/40 [00:03<00:07, 3.84it/s]\n 35%|███▌ | 14/40 [00:03<00:06, 3.84it/s]\n 38%|███▊ | 15/40 [00:03<00:06, 3.84it/s]\n 40%|████ | 16/40 [00:04<00:06, 3.84it/s]\n 42%|████▎ | 17/40 [00:04<00:05, 3.84it/s]\n 45%|████▌ | 18/40 [00:04<00:05, 3.84it/s]\n 48%|████▊ | 19/40 [00:04<00:05, 3.84it/s]\n 50%|█████ | 20/40 [00:05<00:05, 3.84it/s]\n 52%|█████▎ | 21/40 [00:05<00:04, 3.84it/s]\n 55%|█████▌ | 22/40 [00:05<00:04, 3.84it/s]\n 57%|█████▊ | 23/40 [00:05<00:04, 3.83it/s]\n 60%|██████ | 24/40 [00:06<00:04, 3.83it/s]\n 62%|██████▎ | 25/40 [00:06<00:03, 3.83it/s]\n 65%|██████▌ | 26/40 [00:06<00:03, 3.83it/s]\n 68%|██████▊ | 27/40 [00:06<00:03, 3.83it/s]\n 70%|███████ | 28/40 [00:07<00:03, 3.83it/s]\n 72%|███████▎ | 29/40 [00:07<00:02, 3.83it/s]\n 75%|███████▌ | 30/40 [00:07<00:02, 3.83it/s]\n 78%|███████▊ | 31/40 [00:08<00:02, 3.83it/s]\n 80%|████████ | 32/40 [00:08<00:02, 3.83it/s]\n 82%|████████▎ | 33/40 [00:08<00:01, 3.83it/s]\n 85%|████████▌ | 34/40 [00:08<00:01, 3.83it/s]\n 88%|████████▊ | 35/40 [00:09<00:01, 3.83it/s]\n 90%|█████████ | 36/40 [00:09<00:01, 3.83it/s]\n 92%|█████████▎| 37/40 [00:09<00:00, 3.83it/s]\n 95%|█████████▌| 38/40 [00:09<00:00, 3.83it/s]\n 98%|█████████▊| 39/40 [00:10<00:00, 3.83it/s]\n100%|██████████| 40/40 [00:10<00:00, 3.83it/s]\n100%|██████████| 40/40 [00:10<00:00, 3.85it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.715787416, "total_time": 10.723212 }, "output": [ "https://replicate.delivery/xezq/WeDHiWqOUuycHat9z29LsC8MO1fYV3MQ8BtOPVgplrhCpZkUA/out-0.webp" ], "started_at": "2025-04-19T18:15:20.284425Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-rmeasbjo6zpiy4g6uexpw2zaq2dbe6enwuynaqnjtwkgpei2tueq", "get": "https://api.replicate.com/v1/predictions/wqqv8j8vtnrme0cp9x1r25ew4m", "cancel": "https://api.replicate.com/v1/predictions/wqqv8j8vtnrme0cp9x1r25ew4m/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inWeights already loaded Loaded LoRAs in 0.03s Using seed: 39695 Prompt: Subject: TOKMAHI’s full face and neck in a close‑up, eyes piercing through layered abstraction, clad in an oversized acid‑wash denim jacket with bold 90s New York graffiti patches Style: avant‑garde deconstructive abstraction Composition: double‑exposure merge of her portrait with jagged geometric glass shards and neon billboards of Times Square, all on a 45° tilt axis, fragmented planes interlocking with street signs Lighting: sharp metallic rim light slicing through diffuse neon glows from marquee lights and subway car LEDs, punctuated by traffic‑light flares Color Palette: stark blacks and whites accented by neon cyan, hot magenta, acid yellow, plus gritty subway‑green and taxi‑cab amber Mood/Atmosphere: dissonant elegance meets urban grit, fractured surrealism in 90s NYC Technical Details: 50 mm tilt‑shift lens simulation, f/2.0 depth of field, 35 mm analog film aesthetic (Kodachrome vibe), fine film grain, slight vignetting, heavy motion blur on shard edges Additional Elements: glitch‑style digital artifacts, swirling smoke ribbons, reflective metal flakes, drifting taxi light trails, paint‑drip graffiti tags --ar 1:1 --v flux1 --q 2 --style raw --chaos 85 [!] txt2img mode 0%| | 0/40 [00:00<?, ?it/s] 2%|▎ | 1/40 [00:00<00:10, 3.82it/s] 5%|▌ | 2/40 [00:00<00:08, 4.33it/s] 8%|▊ | 3/40 [00:00<00:09, 4.09it/s] 10%|█ | 4/40 [00:00<00:09, 3.99it/s] 12%|█▎ | 5/40 [00:01<00:08, 3.93it/s] 15%|█▌ | 6/40 [00:01<00:08, 3.90it/s] 18%|█▊ | 7/40 [00:01<00:08, 3.88it/s] 20%|██ | 8/40 [00:02<00:08, 3.86it/s] 22%|██▎ | 9/40 [00:02<00:08, 3.86it/s] 25%|██▌ | 10/40 [00:02<00:07, 3.85it/s] 28%|██▊ | 11/40 [00:02<00:07, 3.85it/s] 30%|███ | 12/40 [00:03<00:07, 3.84it/s] 32%|███▎ | 13/40 [00:03<00:07, 3.84it/s] 35%|███▌ | 14/40 [00:03<00:06, 3.84it/s] 38%|███▊ | 15/40 [00:03<00:06, 3.84it/s] 40%|████ | 16/40 [00:04<00:06, 3.84it/s] 42%|████▎ | 17/40 [00:04<00:05, 3.84it/s] 45%|████▌ | 18/40 [00:04<00:05, 3.84it/s] 48%|████▊ | 19/40 [00:04<00:05, 3.84it/s] 50%|█████ | 20/40 [00:05<00:05, 3.84it/s] 52%|█████▎ | 21/40 [00:05<00:04, 3.84it/s] 55%|█████▌ | 22/40 [00:05<00:04, 3.84it/s] 57%|█████▊ | 23/40 [00:05<00:04, 3.83it/s] 60%|██████ | 24/40 [00:06<00:04, 3.83it/s] 62%|██████▎ | 25/40 [00:06<00:03, 3.83it/s] 65%|██████▌ | 26/40 [00:06<00:03, 3.83it/s] 68%|██████▊ | 27/40 [00:06<00:03, 3.83it/s] 70%|███████ | 28/40 [00:07<00:03, 3.83it/s] 72%|███████▎ | 29/40 [00:07<00:02, 3.83it/s] 75%|███████▌ | 30/40 [00:07<00:02, 3.83it/s] 78%|███████▊ | 31/40 [00:08<00:02, 3.83it/s] 80%|████████ | 32/40 [00:08<00:02, 3.83it/s] 82%|████████▎ | 33/40 [00:08<00:01, 3.83it/s] 85%|████████▌ | 34/40 [00:08<00:01, 3.83it/s] 88%|████████▊ | 35/40 [00:09<00:01, 3.83it/s] 90%|█████████ | 36/40 [00:09<00:01, 3.83it/s] 92%|█████████▎| 37/40 [00:09<00:00, 3.83it/s] 95%|█████████▌| 38/40 [00:09<00:00, 3.83it/s] 98%|█████████▊| 39/40 [00:10<00:00, 3.83it/s] 100%|██████████| 40/40 [00:10<00:00, 3.83it/s] 100%|██████████| 40/40 [00:10<00:00, 3.85it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDz018kkgxc9rme0cp9xb8hbfvmrStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines Style: avant-garde deconstructive elegance Composition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth Lighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows Color Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones Mood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit Technical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds Additional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris --ar 3:4 --v flux1 --q 2 --style raw --chaos 80
- go_fast
- lora_scale
- 1.104
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.5
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 16
{ "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 16 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", go_fast: false, lora_scale: 1.104, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.5, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 16 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": False, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 16 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 16 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.104' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.5' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=16'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 16 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:36:13.000311Z", "created_at": "2025-04-19T18:36:05.858000Z", "data_removed": false, "error": null, "id": "z018kkgxc9rme0cp9xb8hbfvmr", "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.5, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 16 }, "logs": "Loaded LoRAs in 2.51s\nUsing seed: 11098\nPrompt: Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines\nStyle: avant-garde deconstructive elegance\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n[!] txt2img mode\n 0%| | 0/16 [00:00<?, ?it/s]\n 6%|▋ | 1/16 [00:00<00:04, 3.73it/s]\n 12%|█▎ | 2/16 [00:00<00:03, 4.13it/s]\n 19%|█▉ | 3/16 [00:00<00:03, 3.98it/s]\n 25%|██▌ | 4/16 [00:01<00:03, 3.92it/s]\n 31%|███▏ | 5/16 [00:01<00:02, 3.89it/s]\n 38%|███▊ | 6/16 [00:01<00:02, 3.85it/s]\n 44%|████▍ | 7/16 [00:01<00:02, 3.84it/s]\n 50%|█████ | 8/16 [00:02<00:02, 3.83it/s]\n 56%|█████▋ | 9/16 [00:02<00:01, 3.83it/s]\n 62%|██████▎ | 10/16 [00:02<00:01, 3.83it/s]\n 69%|██████▉ | 11/16 [00:02<00:01, 3.82it/s]\n 75%|███████▌ | 12/16 [00:03<00:01, 3.82it/s]\n 81%|████████▏ | 13/16 [00:03<00:00, 3.82it/s]\n 88%|████████▊ | 14/16 [00:03<00:00, 3.82it/s]\n 94%|█████████▍| 15/16 [00:03<00:00, 3.82it/s]\n100%|██████████| 16/16 [00:04<00:00, 3.82it/s]\n100%|██████████| 16/16 [00:04<00:00, 3.85it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 6.961592712, "total_time": 7.142311 }, "output": [ "https://replicate.delivery/xezq/ThTO2wNaC5aSFtdEWejcWxcqaTqXuvYUhm5ds0oubUAOeZkUA/out-0.webp" ], "started_at": "2025-04-19T18:36:06.038719Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-h2rzy325iiq5isqql2ftvnsea66at5wa2olebwwf4nl7zzt2kl5q", "get": "https://api.replicate.com/v1/predictions/z018kkgxc9rme0cp9xb8hbfvmr", "cancel": "https://api.replicate.com/v1/predictions/z018kkgxc9rme0cp9xb8hbfvmr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 2.51s Using seed: 11098 Prompt: Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines Style: avant-garde deconstructive elegance Composition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth Lighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows Color Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones Mood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit Technical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds Additional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris --ar 3:4 --v flux1 --q 2 --style raw --chaos 80 [!] txt2img mode 0%| | 0/16 [00:00<?, ?it/s] 6%|▋ | 1/16 [00:00<00:04, 3.73it/s] 12%|█▎ | 2/16 [00:00<00:03, 4.13it/s] 19%|█▉ | 3/16 [00:00<00:03, 3.98it/s] 25%|██▌ | 4/16 [00:01<00:03, 3.92it/s] 31%|███▏ | 5/16 [00:01<00:02, 3.89it/s] 38%|███▊ | 6/16 [00:01<00:02, 3.85it/s] 44%|████▍ | 7/16 [00:01<00:02, 3.84it/s] 50%|█████ | 8/16 [00:02<00:02, 3.83it/s] 56%|█████▋ | 9/16 [00:02<00:01, 3.83it/s] 62%|██████▎ | 10/16 [00:02<00:01, 3.83it/s] 69%|██████▉ | 11/16 [00:02<00:01, 3.82it/s] 75%|███████▌ | 12/16 [00:03<00:01, 3.82it/s] 81%|████████▏ | 13/16 [00:03<00:00, 3.82it/s] 88%|████████▊ | 14/16 [00:03<00:00, 3.82it/s] 94%|█████████▍| 15/16 [00:03<00:00, 3.82it/s] 100%|██████████| 16/16 [00:04<00:00, 3.82it/s] 100%|██████████| 16/16 [00:04<00:00, 3.85it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID6a2nv3v5dhrm80cp9xn8380e6cStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light Style: chiaroscuro Composition: tight close‑up, centered, with half the face in sharp light and half in deep shadow Lighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours Color Palette: pure blacks and whites with subtle mid‑tone grays Mood/Atmosphere: intense, sculptural drama Technical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain Additional Elements: soft edge falloff, barely visible shadow texture on skin --ar 1:1 --v flux1 --q 2 --style raw --chaos 20
- go_fast
- lora_scale
- 1.02
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \nStyle: chiaroscuro \nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \nColor Palette: pure blacks and whites with subtle mid‑tone grays \nMood/Atmosphere: intense, sculptural drama \nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \nStyle: chiaroscuro \nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \nColor Palette: pure blacks and whites with subtle mid‑tone grays \nMood/Atmosphere: intense, sculptural drama \nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", go_fast: false, lora_scale: 1.02, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \nStyle: chiaroscuro \nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \nColor Palette: pure blacks and whites with subtle mid‑tone grays \nMood/Atmosphere: intense, sculptural drama \nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": False, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \\nStyle: chiaroscuro \\nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \\nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \\nColor Palette: pure blacks and whites with subtle mid‑tone grays \\nMood/Atmosphere: intense, sculptural drama \\nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \\nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \\nStyle: chiaroscuro \\nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \\nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \\nColor Palette: pure blacks and whites with subtle mid‑tone grays \\nMood/Atmosphere: intense, sculptural drama \\nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \\nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.02' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \\nStyle: chiaroscuro \\nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \\nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \\nColor Palette: pure blacks and whites with subtle mid‑tone grays \\nMood/Atmosphere: intense, sculptural drama \\nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \\nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:58:24.513438Z", "created_at": "2025-04-19T18:58:15.020000Z", "data_removed": false, "error": null, "id": "6a2nv3v5dhrm80cp9xn8380e6c", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light \nStyle: chiaroscuro \nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow \nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours \nColor Palette: pure blacks and whites with subtle mid‑tone grays \nMood/Atmosphere: intense, sculptural drama \nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain \nAdditional Elements: soft edge falloff, barely visible shadow texture on skin \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.02, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 25379\nPrompt:\nSubject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light\nStyle: chiaroscuro\nComposition: tight close‑up, centered, with half the face in sharp light and half in deep shadow\nLighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours\nColor Palette: pure blacks and whites with subtle mid‑tone grays\nMood/Atmosphere: intense, sculptural drama\nTechnical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain\nAdditional Elements: soft edge falloff, barely visible shadow texture on skin\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.74it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.15it/s]\n 10%|█ | 3/30 [00:00<00:06, 4.00it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.86it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.86it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.85it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.84it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.84it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.84it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.83it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.83it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.82it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.82it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.83it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s]\n 77%|███████▋ | 23/30 [00:05<00:01, 3.83it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.82it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.83it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.83it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.635915543, "total_time": 9.493438 }, "output": [ "https://replicate.delivery/xezq/PbPne9J6F1w6BKt6WT6gbITVwlEYdeAaPBcsDwhJa8hQRakUA/out-0.webp" ], "started_at": "2025-04-19T18:58:15.877522Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-4y7gjfkdottxloht443jufu26ocnbq23xs6iw7ldyulkp3lkbgcq", "get": "https://api.replicate.com/v1/predictions/6a2nv3v5dhrm80cp9xn8380e6c", "cancel": "https://api.replicate.com/v1/predictions/6a2nv3v5dhrm80cp9xn8380e6c/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 25379 Prompt: Subject: TOKMAHI’s face and neck filling the frame, features sculpted by deep shadow and light Style: chiaroscuro Composition: tight close‑up, centered, with half the face in sharp light and half in deep shadow Lighting: single hard light source at a 45° angle, creating dramatic contrast and pronounced shadow contours Color Palette: pure blacks and whites with subtle mid‑tone grays Mood/Atmosphere: intense, sculptural drama Technical Details: 85 mm lens simulation, f/1.4 depth of field, high contrast, fine film grain Additional Elements: soft edge falloff, barely visible shadow texture on skin --ar 1:1 --v flux1 --q 2 --style raw --chaos 20 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.74it/s] 7%|▋ | 2/30 [00:00<00:06, 4.15it/s] 10%|█ | 3/30 [00:00<00:06, 4.00it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.92it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.89it/s] 20%|██ | 6/30 [00:01<00:06, 3.86it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.86it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.85it/s] 30%|███ | 9/30 [00:02<00:05, 3.84it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.84it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.84it/s] 40%|████ | 12/30 [00:03<00:04, 3.84it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.82it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.83it/s] 50%|█████ | 15/30 [00:03<00:03, 3.83it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.82it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.82it/s] 60%|██████ | 18/30 [00:04<00:03, 3.82it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.82it/s] 70%|███████ | 21/30 [00:05<00:02, 3.83it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s] 77%|███████▋ | 23/30 [00:05<00:01, 3.83it/s] 80%|████████ | 24/30 [00:06<00:01, 3.82it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.83it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.83it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.83it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.83it/s] 100%|██████████| 30/30 [00:07<00:00, 3.83it/s] 100%|██████████| 30/30 [00:07<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID62f2edmw2xrma0cp9xrtt2wccrStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and neck in bright, even illumination Style: high‑key beauty Composition: close‑up, centered, minimal shadows Lighting: broad softbox‑style light from front and fill from below Color Palette: clean whites and pastel accents—blush pinks, baby blues Mood/Atmosphere: fresh, airy, flawless luminosity Technical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure Additional Elements: dewy skin finish, soft bokeh background --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 28
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck in bright, even illumination \nStyle: high‑key beauty \nComposition: close‑up, centered, minimal shadows \nLighting: broad softbox‑style light from front and fill from below \nColor Palette: clean whites and pastel accents—blush pinks, baby blues \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \nAdditional Elements: dewy skin finish, soft bokeh background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 28 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and neck in bright, even illumination \nStyle: high‑key beauty \nComposition: close‑up, centered, minimal shadows \nLighting: broad softbox‑style light from front and fill from below \nColor Palette: clean whites and pastel accents—blush pinks, baby blues \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \nAdditional Elements: dewy skin finish, soft bokeh background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 28 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck in bright, even illumination \nStyle: high‑key beauty \nComposition: close‑up, centered, minimal shadows \nLighting: broad softbox‑style light from front and fill from below \nColor Palette: clean whites and pastel accents—blush pinks, baby blues \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \nAdditional Elements: dewy skin finish, soft bokeh background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 28 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and neck in bright, even illumination \\nStyle: high‑key beauty \\nComposition: close‑up, centered, minimal shadows \\nLighting: broad softbox‑style light from front and fill from below \\nColor Palette: clean whites and pastel accents—blush pinks, baby blues \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \\nAdditional Elements: dewy skin finish, soft bokeh background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 28 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and neck in bright, even illumination \\nStyle: high‑key beauty \\nComposition: close‑up, centered, minimal shadows \\nLighting: broad softbox‑style light from front and fill from below \\nColor Palette: clean whites and pastel accents—blush pinks, baby blues \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \\nAdditional Elements: dewy skin finish, soft bokeh background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=28'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and neck in bright, even illumination \\nStyle: high‑key beauty \\nComposition: close‑up, centered, minimal shadows \\nLighting: broad softbox‑style light from front and fill from below \\nColor Palette: clean whites and pastel accents—blush pinks, baby blues \\nMood/Atmosphere: fresh, airy, flawless luminosity \\nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \\nAdditional Elements: dewy skin finish, soft bokeh background \\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 28 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:06:16.140669Z", "created_at": "2025-04-19T19:06:07.767000Z", "data_removed": false, "error": null, "id": "62f2edmw2xrma0cp9xrtt2wccr", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and neck in bright, even illumination \nStyle: high‑key beauty \nComposition: close‑up, centered, minimal shadows \nLighting: broad softbox‑style light from front and fill from below \nColor Palette: clean whites and pastel accents—blush pinks, baby blues \nMood/Atmosphere: fresh, airy, flawless luminosity \nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure \nAdditional Elements: dewy skin finish, soft bokeh background \n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 28 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 27263\nPrompt:\nSubject: TOKMAHI’s face and neck in bright, even illumination\nStyle: high‑key beauty\nComposition: close‑up, centered, minimal shadows\nLighting: broad softbox‑style light from front and fill from below\nColor Palette: clean whites and pastel accents—blush pinks, baby blues\nMood/Atmosphere: fresh, airy, flawless luminosity\nTechnical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure\nAdditional Elements: dewy skin finish, soft bokeh background\n--ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20\n[!] txt2img mode\n 0%| | 0/28 [00:00<?, ?it/s]\n 4%|▎ | 1/28 [00:00<00:07, 3.73it/s]\n 7%|▋ | 2/28 [00:00<00:06, 4.10it/s]\n 11%|█ | 3/28 [00:00<00:06, 3.96it/s]\n 14%|█▍ | 4/28 [00:01<00:06, 3.90it/s]\n 18%|█▊ | 5/28 [00:01<00:05, 3.85it/s]\n 21%|██▏ | 6/28 [00:01<00:05, 3.83it/s]\n 25%|██▌ | 7/28 [00:01<00:05, 3.82it/s]\n 29%|██▊ | 8/28 [00:02<00:05, 3.81it/s]\n 32%|███▏ | 9/28 [00:02<00:04, 3.81it/s]\n 36%|███▌ | 10/28 [00:02<00:04, 3.80it/s]\n 39%|███▉ | 11/28 [00:02<00:04, 3.80it/s]\n 43%|████▎ | 12/28 [00:03<00:04, 3.80it/s]\n 46%|████▋ | 13/28 [00:03<00:03, 3.80it/s]\n 50%|█████ | 14/28 [00:03<00:03, 3.79it/s]\n 54%|█████▎ | 15/28 [00:03<00:03, 3.79it/s]\n 57%|█████▋ | 16/28 [00:04<00:03, 3.79it/s]\n 61%|██████ | 17/28 [00:04<00:02, 3.79it/s]\n 64%|██████▍ | 18/28 [00:04<00:02, 3.79it/s]\n 68%|██████▊ | 19/28 [00:04<00:02, 3.79it/s]\n 71%|███████▏ | 20/28 [00:05<00:02, 3.79it/s]\n 75%|███████▌ | 21/28 [00:05<00:01, 3.79it/s]\n 79%|███████▊ | 22/28 [00:05<00:01, 3.79it/s]\n 82%|████████▏ | 23/28 [00:06<00:01, 3.79it/s]\n 86%|████████▌ | 24/28 [00:06<00:01, 3.79it/s]\n 89%|████████▉ | 25/28 [00:06<00:00, 3.79it/s]\n 93%|█████████▎| 26/28 [00:06<00:00, 3.79it/s]\n 96%|█████████▋| 27/28 [00:07<00:00, 3.79it/s]\n100%|██████████| 28/28 [00:07<00:00, 3.79it/s]\n100%|██████████| 28/28 [00:07<00:00, 3.81it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.207716839, "total_time": 8.373669 }, "output": [ "https://replicate.delivery/xezq/3AOMv8qsRTooBl4aFri6ASJmR5ZBjZ3FlIkXAe669ObUMNSKA/out-0.webp" ], "started_at": "2025-04-19T19:06:07.932952Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-mwnp7v3njisap34zpb3zqh47weaidwkslc76anyc4eohccka5k2a", "get": "https://api.replicate.com/v1/predictions/62f2edmw2xrma0cp9xrtt2wccr", "cancel": "https://api.replicate.com/v1/predictions/62f2edmw2xrma0cp9xrtt2wccr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 27263 Prompt: Subject: TOKMAHI’s face and neck in bright, even illumination Style: high‑key beauty Composition: close‑up, centered, minimal shadows Lighting: broad softbox‑style light from front and fill from below Color Palette: clean whites and pastel accents—blush pinks, baby blues Mood/Atmosphere: fresh, airy, flawless luminosity Technical Details: 100 mm macro lens simulation, f/2.0, zero shadows, high key exposure Additional Elements: dewy skin finish, soft bokeh background --ar 1:1 --v flux1 --q 2 --style cinematic --chaos 20 [!] txt2img mode 0%| | 0/28 [00:00<?, ?it/s] 4%|▎ | 1/28 [00:00<00:07, 3.73it/s] 7%|▋ | 2/28 [00:00<00:06, 4.10it/s] 11%|█ | 3/28 [00:00<00:06, 3.96it/s] 14%|█▍ | 4/28 [00:01<00:06, 3.90it/s] 18%|█▊ | 5/28 [00:01<00:05, 3.85it/s] 21%|██▏ | 6/28 [00:01<00:05, 3.83it/s] 25%|██▌ | 7/28 [00:01<00:05, 3.82it/s] 29%|██▊ | 8/28 [00:02<00:05, 3.81it/s] 32%|███▏ | 9/28 [00:02<00:04, 3.81it/s] 36%|███▌ | 10/28 [00:02<00:04, 3.80it/s] 39%|███▉ | 11/28 [00:02<00:04, 3.80it/s] 43%|████▎ | 12/28 [00:03<00:04, 3.80it/s] 46%|████▋ | 13/28 [00:03<00:03, 3.80it/s] 50%|█████ | 14/28 [00:03<00:03, 3.79it/s] 54%|█████▎ | 15/28 [00:03<00:03, 3.79it/s] 57%|█████▋ | 16/28 [00:04<00:03, 3.79it/s] 61%|██████ | 17/28 [00:04<00:02, 3.79it/s] 64%|██████▍ | 18/28 [00:04<00:02, 3.79it/s] 68%|██████▊ | 19/28 [00:04<00:02, 3.79it/s] 71%|███████▏ | 20/28 [00:05<00:02, 3.79it/s] 75%|███████▌ | 21/28 [00:05<00:01, 3.79it/s] 79%|███████▊ | 22/28 [00:05<00:01, 3.79it/s] 82%|████████▏ | 23/28 [00:06<00:01, 3.79it/s] 86%|████████▌ | 24/28 [00:06<00:01, 3.79it/s] 89%|████████▉ | 25/28 [00:06<00:00, 3.79it/s] 93%|█████████▎| 26/28 [00:06<00:00, 3.79it/s] 96%|█████████▋| 27/28 [00:07<00:00, 3.79it/s] 100%|██████████| 28/28 [00:07<00:00, 3.79it/s] 100%|██████████| 28/28 [00:07<00:00, 3.81it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDm03ptkzrgnrmc0cp9y9bqwsaywStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.22
- output_quality
- 80
- prompt_strength
- 0.5
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "image": "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", model: "dev", prompt: "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.22, output_quality: 80, prompt_strength: 0.5, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.22' \ -i 'output_quality=80' \ -i 'prompt_strength=0.5' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:42:45.404917Z", "created_at": "2025-04-19T19:42:34.117000Z", "data_removed": false, "error": null, "id": "m03ptkzrgnrmc0cp9y9bqwsayw", "input": { "image": "https://replicate.delivery/pbxt/MrjhzrpjG5WU4UzWT5jHDrrtcTHET9IeWuSF66xr8ZJrW0Iy/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_42_20%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 13119\nPrompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue\nStyle: vintage film\nComposition: 3/4 crop from shoulders up, slight head tilt\nLighting: warm golden‑hour glow with gentle side fill\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia\nMood/Atmosphere: nostalgic, timeless warmth\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette\nAdditional Elements: subtle light leaks, soft border fade\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:04, 3.42it/s]\n 13%|█▎ | 2/15 [00:00<00:04, 3.09it/s]\n 20%|██ | 3/15 [00:00<00:04, 3.00it/s]\n 27%|██▋ | 4/15 [00:01<00:03, 2.96it/s]\n 33%|███▎ | 5/15 [00:01<00:03, 2.93it/s]\n 40%|████ | 6/15 [00:02<00:03, 2.92it/s]\n 47%|████▋ | 7/15 [00:02<00:02, 2.91it/s]\n 53%|█████▎ | 8/15 [00:02<00:02, 2.90it/s]\n 60%|██████ | 9/15 [00:03<00:02, 2.90it/s]\n 67%|██████▋ | 10/15 [00:03<00:01, 2.90it/s]\n 73%|███████▎ | 11/15 [00:03<00:01, 2.90it/s]\n 80%|████████ | 12/15 [00:04<00:01, 2.90it/s]\n 87%|████████▋ | 13/15 [00:04<00:00, 2.90it/s]\n 93%|█████████▎| 14/15 [00:04<00:00, 2.90it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.90it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.92it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 6.336856228, "total_time": 11.287917 }, "output": [ "https://replicate.delivery/xezq/ak5OaebWemme7pJx9gWE4kOrqG2Q6NWH4ILTKo5MTmqq11IpA/out-0.webp" ], "started_at": "2025-04-19T19:42:39.068061Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-lfj5yttk5zgbdiu3f4qm7uradrra2wrsvykxhscu6ltxvs5adk3a", "get": "https://api.replicate.com/v1/predictions/m03ptkzrgnrmc0cp9y9bqwsayw", "cancel": "https://api.replicate.com/v1/predictions/m03ptkzrgnrmc0cp9y9bqwsayw/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 13119 Prompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20 Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:04, 3.42it/s] 13%|█▎ | 2/15 [00:00<00:04, 3.09it/s] 20%|██ | 3/15 [00:00<00:04, 3.00it/s] 27%|██▋ | 4/15 [00:01<00:03, 2.96it/s] 33%|███▎ | 5/15 [00:01<00:03, 2.93it/s] 40%|████ | 6/15 [00:02<00:03, 2.92it/s] 47%|████▋ | 7/15 [00:02<00:02, 2.91it/s] 53%|█████▎ | 8/15 [00:02<00:02, 2.90it/s] 60%|██████ | 9/15 [00:03<00:02, 2.90it/s] 67%|██████▋ | 10/15 [00:03<00:01, 2.90it/s] 73%|███████▎ | 11/15 [00:03<00:01, 2.90it/s] 80%|████████ | 12/15 [00:04<00:01, 2.90it/s] 87%|████████▋ | 13/15 [00:04<00:00, 2.90it/s] 93%|█████████▎| 14/15 [00:04<00:00, 2.90it/s] 100%|██████████| 15/15 [00:05<00:00, 2.90it/s] 100%|██████████| 15/15 [00:05<00:00, 2.92it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDyg4c9yzdhxrmc0cp9y2s2jqyh0StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief Style: surreal macro abstraction Composition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off Lighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals Color Palette: natural skin tones with electric blue iris accent and translucent white petals Mood/Atmosphere: uncanny beauty, hyper‑real intensity Technical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain Additional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.22
- output_quality
- 80
- prompt_strength
- 0.5
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "image": "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \nStyle: surreal macro abstraction \nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \nMood/Atmosphere: uncanny beauty, hyper‑real intensity \nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", model: "dev", prompt: "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \nStyle: surreal macro abstraction \nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \nMood/Atmosphere: uncanny beauty, hyper‑real intensity \nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.22, output_quality: 80, prompt_strength: 0.5, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \nStyle: surreal macro abstraction \nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \nMood/Atmosphere: uncanny beauty, hyper‑real intensity \nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \\nStyle: surreal macro abstraction \\nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \\nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \\nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \\nMood/Atmosphere: uncanny beauty, hyper‑real intensity \\nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \\nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \\nStyle: surreal macro abstraction \\nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \\nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \\nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \\nMood/Atmosphere: uncanny beauty, hyper‑real intensity \\nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \\nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 "' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.22' \ -i 'output_quality=80' \ -i 'prompt_strength=0.5' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \\nStyle: surreal macro abstraction \\nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \\nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \\nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \\nMood/Atmosphere: uncanny beauty, hyper‑real intensity \\nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \\nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:28:25.847050Z", "created_at": "2025-04-19T19:28:19.343000Z", "data_removed": false, "error": null, "id": "yg4c9yzdhxrmc0cp9y2s2jqyh0", "input": { "image": "https://replicate.delivery/pbxt/MrjUOkzjz2gNWbHfdPbRAzFV3lD4ltrQyOjCuYb7VXmwXNYE/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_26_44%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief \nStyle: surreal macro abstraction \nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off \nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals \nColor Palette: natural skin tones with electric blue iris accent and translucent white petals \nMood/Atmosphere: uncanny beauty, hyper‑real intensity \nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain \nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 ", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.59s\nUsing seed: 4190\nPrompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief\nStyle: surreal macro abstraction\nComposition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off\nLighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals\nColor Palette: natural skin tones with electric blue iris accent and translucent white petals\nMood/Atmosphere: uncanny beauty, hyper‑real intensity\nTechnical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain\nAdditional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60\nInput image size: 1024x1536\n[!] Resizing input image from 1024x1536 to 960x1440\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:04, 3.43it/s]\n 13%|█▎ | 2/15 [00:00<00:04, 3.10it/s]\n 20%|██ | 3/15 [00:00<00:04, 3.00it/s]\n 27%|██▋ | 4/15 [00:01<00:03, 2.94it/s]\n 33%|███▎ | 5/15 [00:01<00:03, 2.93it/s]\n 40%|████ | 6/15 [00:02<00:03, 2.91it/s]\n 47%|████▋ | 7/15 [00:02<00:02, 2.91it/s]\n 53%|█████▎ | 8/15 [00:02<00:02, 2.90it/s]\n 60%|██████ | 9/15 [00:03<00:02, 2.89it/s]\n 67%|██████▋ | 10/15 [00:03<00:01, 2.88it/s]\n 73%|███████▎ | 11/15 [00:03<00:01, 2.89it/s]\n 80%|████████ | 12/15 [00:04<00:01, 2.88it/s]\n 87%|████████▋ | 13/15 [00:04<00:00, 2.88it/s]\n 93%|█████████▎| 14/15 [00:04<00:00, 2.88it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.88it/s]\n100%|██████████| 15/15 [00:05<00:00, 2.91it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 6.393578867, "total_time": 6.50405 }, "output": [ "https://replicate.delivery/xezq/eYXWKLoYarXOdyE5PiF34aHh2bJpe0ppfhDVqSluFDLya1IpA/out-0.webp" ], "started_at": "2025-04-19T19:28:19.453471Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-xzk4hwrvqwt42whj4bnjcl7fwfaxsgyxqh3mnsjsvwbvnsyovs6a", "get": "https://api.replicate.com/v1/predictions/yg4c9yzdhxrmc0cp9y2s2jqyh0", "cancel": "https://api.replicate.com/v1/predictions/yg4c9yzdhxrmc0cp9y2s2jqyh0/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.59s Using seed: 4190 Prompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, every pore and micro‑gloss droplet in razor‑sharp relief Style: surreal macro abstraction Composition: extreme tight crop, centered—100 mm macro lens simulation, f/2.8, seamless focus fall‑off Lighting: LED ring‑light catch‑dots in eyes, soft backlight to silhouette resin petals Color Palette: natural skin tones with electric blue iris accent and translucent white petals Mood/Atmosphere: uncanny beauty, hyper‑real intensity Technical Details: micro‑detailed skin texture, slight motion blur on stray hairs, fine film grain Additional Elements: tiny floating water‑droplet motes morphing into miniature galaxies, faint lens flare arcs --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 60 Input image size: 1024x1536 [!] Resizing input image from 1024x1536 to 960x1440 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:04, 3.43it/s] 13%|█▎ | 2/15 [00:00<00:04, 3.10it/s] 20%|██ | 3/15 [00:00<00:04, 3.00it/s] 27%|██▋ | 4/15 [00:01<00:03, 2.94it/s] 33%|███▎ | 5/15 [00:01<00:03, 2.93it/s] 40%|████ | 6/15 [00:02<00:03, 2.91it/s] 47%|████▋ | 7/15 [00:02<00:02, 2.91it/s] 53%|█████▎ | 8/15 [00:02<00:02, 2.90it/s] 60%|██████ | 9/15 [00:03<00:02, 2.89it/s] 67%|██████▋ | 10/15 [00:03<00:01, 2.88it/s] 73%|███████▎ | 11/15 [00:03<00:01, 2.89it/s] 80%|████████ | 12/15 [00:04<00:01, 2.88it/s] 87%|████████▋ | 13/15 [00:04<00:00, 2.88it/s] 93%|█████████▎| 14/15 [00:04<00:00, 2.88it/s] 100%|██████████| 15/15 [00:05<00:00, 2.88it/s] 100%|██████████| 15/15 [00:05<00:00, 2.91it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID6hak9s7xenrmc0cp9vfad4xbngStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence Style: enchanted surrealism Composition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light Lighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair Color Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white Mood/Atmosphere: mystical, intimate, otherworldly serenity Technical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps Additional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.01
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 50
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.01, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 50 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.01' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=50'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \\nStyle: enchanted surrealism \\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \\nMood/Atmosphere: mystical, intimate, otherworldly serenity \\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T16:26:16.544563Z", "created_at": "2025-04-19T16:25:58.901000Z", "data_removed": false, "error": null, "id": "6hak9s7xenrmc0cp9vfad4xbng", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence \nStyle: enchanted surrealism \nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light \nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair \nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white \nMood/Atmosphere: mystical, intimate, otherworldly serenity \nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps \nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }, "logs": "Loaded LoRAs in 0.56s\nUsing seed: 41115\nPrompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence\nStyle: enchanted surrealism\nComposition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light\nLighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair\nColor Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white\nMood/Atmosphere: mystical, intimate, otherworldly serenity\nTechnical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps\nAdditional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop\n[!] txt2img mode\n 0%| | 0/50 [00:00<?, ?it/s]\n 2%|▏ | 1/50 [00:00<00:13, 3.72it/s]\n 4%|▍ | 2/50 [00:00<00:11, 4.10it/s]\n 6%|▌ | 3/50 [00:00<00:11, 3.97it/s]\n 8%|▊ | 4/50 [00:01<00:11, 3.90it/s]\n 10%|█ | 5/50 [00:01<00:11, 3.86it/s]\n 12%|█▏ | 6/50 [00:01<00:11, 3.83it/s]\n 14%|█▍ | 7/50 [00:01<00:11, 3.82it/s]\n 16%|█▌ | 8/50 [00:02<00:11, 3.82it/s]\n 18%|█▊ | 9/50 [00:02<00:10, 3.81it/s]\n 20%|██ | 10/50 [00:02<00:10, 3.80it/s]\n 22%|██▏ | 11/50 [00:02<00:10, 3.81it/s]\n 24%|██▍ | 12/50 [00:03<00:09, 3.81it/s]\n 26%|██▌ | 13/50 [00:03<00:09, 3.80it/s]\n 28%|██▊ | 14/50 [00:03<00:09, 3.80it/s]\n 30%|███ | 15/50 [00:03<00:09, 3.80it/s]\n 32%|███▏ | 16/50 [00:04<00:08, 3.80it/s]\n 34%|███▍ | 17/50 [00:04<00:08, 3.79it/s]\n 36%|███▌ | 18/50 [00:04<00:08, 3.80it/s]\n 38%|███▊ | 19/50 [00:04<00:08, 3.80it/s]\n 40%|████ | 20/50 [00:05<00:07, 3.80it/s]\n 42%|████▏ | 21/50 [00:05<00:07, 3.80it/s]\n 44%|████▍ | 22/50 [00:05<00:07, 3.80it/s]\n 46%|████▌ | 23/50 [00:06<00:07, 3.80it/s]\n 48%|████▊ | 24/50 [00:06<00:06, 3.79it/s]\n 50%|█████ | 25/50 [00:06<00:06, 3.80it/s]\n 52%|█████▏ | 26/50 [00:06<00:06, 3.80it/s]\n 54%|█████▍ | 27/50 [00:07<00:06, 3.80it/s]\n 56%|█████▌ | 28/50 [00:07<00:05, 3.80it/s]\n 58%|█████▊ | 29/50 [00:07<00:05, 3.80it/s]\n 60%|██████ | 30/50 [00:07<00:05, 3.80it/s]\n 62%|██████▏ | 31/50 [00:08<00:05, 3.80it/s]\n 64%|██████▍ | 32/50 [00:08<00:04, 3.80it/s]\n 66%|██████▌ | 33/50 [00:08<00:04, 3.80it/s]\n 68%|██████▊ | 34/50 [00:08<00:04, 3.80it/s]\n 70%|███████ | 35/50 [00:09<00:03, 3.79it/s]\n 72%|███████▏ | 36/50 [00:09<00:03, 3.79it/s]\n 74%|███████▍ | 37/50 [00:09<00:03, 3.79it/s]\n 76%|███████▌ | 38/50 [00:09<00:03, 3.80it/s]\n 78%|███████▊ | 39/50 [00:10<00:02, 3.79it/s]\n 80%|████████ | 40/50 [00:10<00:02, 3.79it/s]\n 82%|████████▏ | 41/50 [00:10<00:02, 3.79it/s]\n 84%|████████▍ | 42/50 [00:11<00:02, 3.80it/s]\n 86%|████████▌ | 43/50 [00:11<00:01, 3.79it/s]\n 88%|████████▊ | 44/50 [00:11<00:01, 3.79it/s]\n 90%|█████████ | 45/50 [00:11<00:01, 3.80it/s]\n 92%|█████████▏| 46/50 [00:12<00:01, 3.80it/s]\n 94%|█████████▍| 47/50 [00:12<00:00, 3.80it/s]\n 96%|█████████▌| 48/50 [00:12<00:00, 3.80it/s]\n 98%|█████████▊| 49/50 [00:12<00:00, 3.80it/s]\n100%|██████████| 50/50 [00:13<00:00, 3.80it/s]\n100%|██████████| 50/50 [00:13<00:00, 3.81it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 13.9799359, "total_time": 17.643563 }, "output": [ "https://replicate.delivery/xezq/HTF2epOiMZSpK6TvTRj7NFQbZw8pe6DSfPghtj8KsHBRFwIpA/out-0.webp" ], "started_at": "2025-04-19T16:26:02.564627Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-d7lt6levpubhdawaekr2iota5ouqmbkpx4fzvuynkrah4cugoupa", "get": "https://api.replicate.com/v1/predictions/6hak9s7xenrmc0cp9vfad4xbng", "cancel": "https://api.replicate.com/v1/predictions/6hak9s7xenrmc0cp9vfad4xbng/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.56s Using seed: 41115 Prompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes gazing calmly ahead, her exaggerated afro sculpted into swirling fungus‑like spirals tipped in glowing bioluminescence Style: enchanted surrealism Composition: tight crop from shoulders up, surrounded by towering, overscaled luminescent mushrooms and ferns that seem to pulse with light Lighting: soft under‑lighting from the bioluminescent flora casting teal and violet glows across her features, subtle rim highlights on her hair Color Palette: deep emerald greens, electric teals, soft violets, and hints of pearl white Mood/Atmosphere: mystical, intimate, otherworldly serenity Technical Details: tack‑sharp focus on her eyes and skin texture; gentle motion blur on the edges of glowing mushroom caps Additional Elements: drifting spores like tiny stars, tendrils of mist weaving between fungal stalks beyond the crop [!] txt2img mode 0%| | 0/50 [00:00<?, ?it/s] 2%|▏ | 1/50 [00:00<00:13, 3.72it/s] 4%|▍ | 2/50 [00:00<00:11, 4.10it/s] 6%|▌ | 3/50 [00:00<00:11, 3.97it/s] 8%|▊ | 4/50 [00:01<00:11, 3.90it/s] 10%|█ | 5/50 [00:01<00:11, 3.86it/s] 12%|█▏ | 6/50 [00:01<00:11, 3.83it/s] 14%|█▍ | 7/50 [00:01<00:11, 3.82it/s] 16%|█▌ | 8/50 [00:02<00:11, 3.82it/s] 18%|█▊ | 9/50 [00:02<00:10, 3.81it/s] 20%|██ | 10/50 [00:02<00:10, 3.80it/s] 22%|██▏ | 11/50 [00:02<00:10, 3.81it/s] 24%|██▍ | 12/50 [00:03<00:09, 3.81it/s] 26%|██▌ | 13/50 [00:03<00:09, 3.80it/s] 28%|██▊ | 14/50 [00:03<00:09, 3.80it/s] 30%|███ | 15/50 [00:03<00:09, 3.80it/s] 32%|███▏ | 16/50 [00:04<00:08, 3.80it/s] 34%|███▍ | 17/50 [00:04<00:08, 3.79it/s] 36%|███▌ | 18/50 [00:04<00:08, 3.80it/s] 38%|███▊ | 19/50 [00:04<00:08, 3.80it/s] 40%|████ | 20/50 [00:05<00:07, 3.80it/s] 42%|████▏ | 21/50 [00:05<00:07, 3.80it/s] 44%|████▍ | 22/50 [00:05<00:07, 3.80it/s] 46%|████▌ | 23/50 [00:06<00:07, 3.80it/s] 48%|████▊ | 24/50 [00:06<00:06, 3.79it/s] 50%|█████ | 25/50 [00:06<00:06, 3.80it/s] 52%|█████▏ | 26/50 [00:06<00:06, 3.80it/s] 54%|█████▍ | 27/50 [00:07<00:06, 3.80it/s] 56%|█████▌ | 28/50 [00:07<00:05, 3.80it/s] 58%|█████▊ | 29/50 [00:07<00:05, 3.80it/s] 60%|██████ | 30/50 [00:07<00:05, 3.80it/s] 62%|██████▏ | 31/50 [00:08<00:05, 3.80it/s] 64%|██████▍ | 32/50 [00:08<00:04, 3.80it/s] 66%|██████▌ | 33/50 [00:08<00:04, 3.80it/s] 68%|██████▊ | 34/50 [00:08<00:04, 3.80it/s] 70%|███████ | 35/50 [00:09<00:03, 3.79it/s] 72%|███████▏ | 36/50 [00:09<00:03, 3.79it/s] 74%|███████▍ | 37/50 [00:09<00:03, 3.79it/s] 76%|███████▌ | 38/50 [00:09<00:03, 3.80it/s] 78%|███████▊ | 39/50 [00:10<00:02, 3.79it/s] 80%|████████ | 40/50 [00:10<00:02, 3.79it/s] 82%|████████▏ | 41/50 [00:10<00:02, 3.79it/s] 84%|████████▍ | 42/50 [00:11<00:02, 3.80it/s] 86%|████████▌ | 43/50 [00:11<00:01, 3.79it/s] 88%|████████▊ | 44/50 [00:11<00:01, 3.79it/s] 90%|█████████ | 45/50 [00:11<00:01, 3.80it/s] 92%|█████████▏| 46/50 [00:12<00:01, 3.80it/s] 94%|█████████▍| 47/50 [00:12<00:00, 3.80it/s] 96%|█████████▌| 48/50 [00:12<00:00, 3.80it/s] 98%|█████████▊| 49/50 [00:12<00:00, 3.80it/s] 100%|██████████| 50/50 [00:13<00:00, 3.80it/s] 100%|██████████| 50/50 [00:13<00:00, 3.81it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDs3bjw35krhrme0cp9vg92cwmm8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct Style: ethereal blend fusion Composition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair Lighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours Color Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas Mood/Atmosphere: dreamlike, transcendent merge of human and nature Technical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements Additional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.01
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 38
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \nStyle: ethereal blend fusion \nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \nMood/Atmosphere: dreamlike, transcendent merge of human and nature \nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \nStyle: ethereal blend fusion \nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \nMood/Atmosphere: dreamlike, transcendent merge of human and nature \nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.01, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 38 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \nStyle: ethereal blend fusion \nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \nMood/Atmosphere: dreamlike, transcendent merge of human and nature \nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \\nStyle: ethereal blend fusion \\nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \\nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \\nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \\nMood/Atmosphere: dreamlike, transcendent merge of human and nature \\nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \\nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \\nStyle: ethereal blend fusion \\nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \\nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \\nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \\nMood/Atmosphere: dreamlike, transcendent merge of human and nature \\nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \\nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.01' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=38'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \\nStyle: ethereal blend fusion \\nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \\nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \\nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \\nMood/Atmosphere: dreamlike, transcendent merge of human and nature \\nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \\nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T16:28:12.261979Z", "created_at": "2025-04-19T16:27:51.108000Z", "data_removed": false, "error": null, "id": "s3bjw35krhrme0cp9vg92cwmm8", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct \nStyle: ethereal blend fusion \nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair \nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours \nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas \nMood/Atmosphere: dreamlike, transcendent merge of human and nature \nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements \nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.01, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 38 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 47489\nPrompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct\nStyle: ethereal blend fusion\nComposition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair\nLighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours\nColor Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas\nMood/Atmosphere: dreamlike, transcendent merge of human and nature\nTechnical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements\nAdditional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/38 [00:00<?, ?it/s]\n 3%|▎ | 1/38 [00:00<00:09, 3.73it/s]\n 5%|▌ | 2/38 [00:00<00:08, 4.14it/s]\n 8%|▊ | 3/38 [00:00<00:08, 3.99it/s]\n 11%|█ | 4/38 [00:01<00:08, 3.91it/s]\n 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s]\n 16%|█▌ | 6/38 [00:01<00:08, 3.87it/s]\n 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s]\n 21%|██ | 8/38 [00:02<00:07, 3.83it/s]\n 24%|██▎ | 9/38 [00:02<00:07, 3.83it/s]\n 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s]\n 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s]\n 32%|███▏ | 12/38 [00:03<00:06, 3.82it/s]\n 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s]\n 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s]\n 39%|███▉ | 15/38 [00:03<00:06, 3.82it/s]\n 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s]\n 45%|████▍ | 17/38 [00:04<00:05, 3.81it/s]\n 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s]\n 50%|█████ | 19/38 [00:04<00:04, 3.82it/s]\n 53%|█████▎ | 20/38 [00:05<00:04, 3.83it/s]\n 55%|█████▌ | 21/38 [00:05<00:04, 3.82it/s]\n 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s]\n 61%|██████ | 23/38 [00:05<00:03, 3.82it/s]\n 63%|██████▎ | 24/38 [00:06<00:03, 3.81it/s]\n 66%|██████▌ | 25/38 [00:06<00:03, 3.81it/s]\n 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s]\n 71%|███████ | 27/38 [00:07<00:02, 3.81it/s]\n 74%|███████▎ | 28/38 [00:07<00:02, 3.82it/s]\n 76%|███████▋ | 29/38 [00:07<00:02, 3.82it/s]\n 79%|███████▉ | 30/38 [00:07<00:02, 3.82it/s]\n 82%|████████▏ | 31/38 [00:08<00:01, 3.82it/s]\n 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s]\n 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s]\n 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s]\n 92%|█████████▏| 35/38 [00:09<00:00, 3.81it/s]\n 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s]\n 97%|█████████▋| 37/38 [00:09<00:00, 3.82it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.82it/s]\n100%|██████████| 38/38 [00:09<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.766772, "total_time": 21.153979 }, "output": [ "https://replicate.delivery/xezq/trZaeP5nFgT0VqsWZeUOj3hihg3WROef1gndh240ui9wRgRSB/out-0.webp" ], "started_at": "2025-04-19T16:28:01.495207Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-gepiyjdhv3dvo32cpbd3tbfm74ujcvmpcabrapfnvogyohk7bopa", "get": "https://api.replicate.com/v1/predictions/s3bjw35krhrme0cp9vg92cwmm8", "cancel": "https://api.replicate.com/v1/predictions/s3bjw35krhrme0cp9vg92cwmm8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 47489 Prompt: Subject: TOKMAHI’s full face and neck filling the frame, eyes serene and direct Style: ethereal blend fusion Composition: double‑exposure blend of her portrait with soft, swirling cloudscapes and delicate floral silhouettes merging into her hair Lighting: diffused pearlescent glow from within, gentle rim highlights tracing her facial contours Color Palette: translucent pastels—misty lilacs, opal pinks, translucent aquas Mood/Atmosphere: dreamlike, transcendent merge of human and nature Technical Details: seamless blend edges, tack‑sharp focus on her eyes and lips, soft motion blur on overlapping elements Additional Elements: faint ghost‑petals drifting across her shoulders, wisps of cloudy texture weaving through her exaggerated afro --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/38 [00:00<?, ?it/s] 3%|▎ | 1/38 [00:00<00:09, 3.73it/s] 5%|▌ | 2/38 [00:00<00:08, 4.14it/s] 8%|▊ | 3/38 [00:00<00:08, 3.99it/s] 11%|█ | 4/38 [00:01<00:08, 3.91it/s] 13%|█▎ | 5/38 [00:01<00:08, 3.89it/s] 16%|█▌ | 6/38 [00:01<00:08, 3.87it/s] 18%|█▊ | 7/38 [00:01<00:08, 3.85it/s] 21%|██ | 8/38 [00:02<00:07, 3.83it/s] 24%|██▎ | 9/38 [00:02<00:07, 3.83it/s] 26%|██▋ | 10/38 [00:02<00:07, 3.83it/s] 29%|██▉ | 11/38 [00:02<00:07, 3.82it/s] 32%|███▏ | 12/38 [00:03<00:06, 3.82it/s] 34%|███▍ | 13/38 [00:03<00:06, 3.82it/s] 37%|███▋ | 14/38 [00:03<00:06, 3.82it/s] 39%|███▉ | 15/38 [00:03<00:06, 3.82it/s] 42%|████▏ | 16/38 [00:04<00:05, 3.82it/s] 45%|████▍ | 17/38 [00:04<00:05, 3.81it/s] 47%|████▋ | 18/38 [00:04<00:05, 3.82it/s] 50%|█████ | 19/38 [00:04<00:04, 3.82it/s] 53%|█████▎ | 20/38 [00:05<00:04, 3.83it/s] 55%|█████▌ | 21/38 [00:05<00:04, 3.82it/s] 58%|█████▊ | 22/38 [00:05<00:04, 3.82it/s] 61%|██████ | 23/38 [00:05<00:03, 3.82it/s] 63%|██████▎ | 24/38 [00:06<00:03, 3.81it/s] 66%|██████▌ | 25/38 [00:06<00:03, 3.81it/s] 68%|██████▊ | 26/38 [00:06<00:03, 3.82it/s] 71%|███████ | 27/38 [00:07<00:02, 3.81it/s] 74%|███████▎ | 28/38 [00:07<00:02, 3.82it/s] 76%|███████▋ | 29/38 [00:07<00:02, 3.82it/s] 79%|███████▉ | 30/38 [00:07<00:02, 3.82it/s] 82%|████████▏ | 31/38 [00:08<00:01, 3.82it/s] 84%|████████▍ | 32/38 [00:08<00:01, 3.81it/s] 87%|████████▋ | 33/38 [00:08<00:01, 3.82it/s] 89%|████████▉ | 34/38 [00:08<00:01, 3.82it/s] 92%|█████████▏| 35/38 [00:09<00:00, 3.81it/s] 95%|█████████▍| 36/38 [00:09<00:00, 3.82it/s] 97%|█████████▋| 37/38 [00:09<00:00, 3.82it/s] 100%|██████████| 38/38 [00:09<00:00, 3.82it/s] 100%|██████████| 38/38 [00:09<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDxxr901skfhrme0cp9vwv4batn8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her Style: regal ethereal fantasy Composition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake Lighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone Color Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips Additional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.78
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 29
{ "model": "dev", "prompt": " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \nStyle: regal ethereal fantasy \nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace \n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.78, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 29 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \nStyle: regal ethereal fantasy \nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace \n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \n", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.78, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 29 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \nStyle: regal ethereal fantasy \nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace \n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \n", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.78, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 29 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \\nStyle: regal ethereal fantasy \\nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \\nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \\nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace \\n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \\nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.78, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 29 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt=" TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \\nStyle: regal ethereal fantasy \\nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \\nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \\nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace \\n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \\nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.78' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=29'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \\nStyle: regal ethereal fantasy \\nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \\nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \\nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace \\n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \\nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.78, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 29 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T16:54:46.729442Z", "created_at": "2025-04-19T16:54:36.668000Z", "data_removed": false, "error": null, "id": "xxr901skfhrme0cp9vwv4batn8", "input": { "model": "dev", "prompt": " TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her \nStyle: regal ethereal fantasy \nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake \nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone \nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace \n tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips \nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 \n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.78, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 29 }, "logs": "free=25579268386816\nDownloading weights\n2025-04-19T16:54:37Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmps9oelp0l/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T16:54:37Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T16:54:37Z | INFO | [ Complete ] dest=/tmp/tmps9oelp0l/weights size=\"172 MB\" total_elapsed=0.516s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 0.53s\nLoaded LoRAs in 1.13s\nUsing seed: 55621\nPrompt: TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her\nStyle: regal ethereal fantasy\nComposition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake\nLighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone\nColor Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue\nMood/Atmosphere: majestic calm, otherworldly grace\ntack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips\nAdditional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/29 [00:00<?, ?it/s]\n 3%|▎ | 1/29 [00:00<00:07, 3.72it/s]\n 7%|▋ | 2/29 [00:00<00:06, 4.10it/s]\n 10%|█ | 3/29 [00:00<00:06, 3.96it/s]\n 14%|█▍ | 4/29 [00:01<00:06, 3.91it/s]\n 17%|█▋ | 5/29 [00:01<00:06, 3.88it/s]\n 21%|██ | 6/29 [00:01<00:05, 3.86it/s]\n 24%|██▍ | 7/29 [00:01<00:05, 3.85it/s]\n 28%|██▊ | 8/29 [00:02<00:05, 3.84it/s]\n 31%|███ | 9/29 [00:02<00:05, 3.83it/s]\n 34%|███▍ | 10/29 [00:02<00:04, 3.83it/s]\n 38%|███▊ | 11/29 [00:02<00:04, 3.83it/s]\n 41%|████▏ | 12/29 [00:03<00:04, 3.83it/s]\n 45%|████▍ | 13/29 [00:03<00:04, 3.82it/s]\n 48%|████▊ | 14/29 [00:03<00:03, 3.82it/s]\n 52%|█████▏ | 15/29 [00:03<00:03, 3.82it/s]\n 55%|█████▌ | 16/29 [00:04<00:03, 3.82it/s]\n 59%|█████▊ | 17/29 [00:04<00:03, 3.82it/s]\n 62%|██████▏ | 18/29 [00:04<00:02, 3.82it/s]\n 66%|██████▌ | 19/29 [00:04<00:02, 3.82it/s]\n 69%|██████▉ | 20/29 [00:05<00:02, 3.81it/s]\n 72%|███████▏ | 21/29 [00:05<00:02, 3.81it/s]\n 76%|███████▌ | 22/29 [00:05<00:01, 3.81it/s]\n 79%|███████▉ | 23/29 [00:05<00:01, 3.82it/s]\n 83%|████████▎ | 24/29 [00:06<00:01, 3.82it/s]\n 86%|████████▌ | 25/29 [00:06<00:01, 3.82it/s]\n 90%|████████▉ | 26/29 [00:06<00:00, 3.83it/s]\n 93%|█████████▎| 27/29 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 28/29 [00:07<00:00, 3.83it/s]\n100%|██████████| 29/29 [00:07<00:00, 3.83it/s]\n100%|██████████| 29/29 [00:07<00:00, 3.84it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.366263765, "total_time": 10.061442 }, "output": [ "https://replicate.delivery/xezq/8VfnbbwI4R2rIieVaGV8RoodhVyZBUsytCOE1Oquue4t6wIpA/out-0.webp" ], "started_at": "2025-04-19T16:54:37.363179Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-qet33loghfhmn4qhu2jnayns2fpuy47qgyuztzdltsjflmbpxxka", "get": "https://api.replicate.com/v1/predictions/xxr901skfhrme0cp9vwv4batn8", "cancel": "https://api.replicate.com/v1/predictions/xxr901skfhrme0cp9vwv4batn8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25579268386816 Downloading weights 2025-04-19T16:54:37Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmps9oelp0l/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T16:54:37Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T16:54:37Z | INFO | [ Complete ] dest=/tmp/tmps9oelp0l/weights size="172 MB" total_elapsed=0.516s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 0.53s Loaded LoRAs in 1.13s Using seed: 55621 Prompt: TOKMAHI lying face‑up, her eyes open in serene majesty, framed by an exaggerated, halo‑like afro that fans out beneath her Style: regal ethereal fantasy Composition: overhead view, tight square crop from chin to shoulders, inverted‑gravity clouds drifting around her as if she’s floating on a celestial lake Lighting: soft divine beams from above casting a warm, golden aura around her head, subtle rim highlights on her hair and collarbone Color Palette: regal purples, soft golds, and pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace tack‑sharp focus on her eyes and face, gentle motion blur on surrounding clouds and hair tips Additional Elements: faint constellations forming a crown pattern in the mist, shimmering motes of light rising like incense smoke --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/29 [00:00<?, ?it/s] 3%|▎ | 1/29 [00:00<00:07, 3.72it/s] 7%|▋ | 2/29 [00:00<00:06, 4.10it/s] 10%|█ | 3/29 [00:00<00:06, 3.96it/s] 14%|█▍ | 4/29 [00:01<00:06, 3.91it/s] 17%|█▋ | 5/29 [00:01<00:06, 3.88it/s] 21%|██ | 6/29 [00:01<00:05, 3.86it/s] 24%|██▍ | 7/29 [00:01<00:05, 3.85it/s] 28%|██▊ | 8/29 [00:02<00:05, 3.84it/s] 31%|███ | 9/29 [00:02<00:05, 3.83it/s] 34%|███▍ | 10/29 [00:02<00:04, 3.83it/s] 38%|███▊ | 11/29 [00:02<00:04, 3.83it/s] 41%|████▏ | 12/29 [00:03<00:04, 3.83it/s] 45%|████▍ | 13/29 [00:03<00:04, 3.82it/s] 48%|████▊ | 14/29 [00:03<00:03, 3.82it/s] 52%|█████▏ | 15/29 [00:03<00:03, 3.82it/s] 55%|█████▌ | 16/29 [00:04<00:03, 3.82it/s] 59%|█████▊ | 17/29 [00:04<00:03, 3.82it/s] 62%|██████▏ | 18/29 [00:04<00:02, 3.82it/s] 66%|██████▌ | 19/29 [00:04<00:02, 3.82it/s] 69%|██████▉ | 20/29 [00:05<00:02, 3.81it/s] 72%|███████▏ | 21/29 [00:05<00:02, 3.81it/s] 76%|███████▌ | 22/29 [00:05<00:01, 3.81it/s] 79%|███████▉ | 23/29 [00:05<00:01, 3.82it/s] 83%|████████▎ | 24/29 [00:06<00:01, 3.82it/s] 86%|████████▌ | 25/29 [00:06<00:01, 3.82it/s] 90%|████████▉ | 26/29 [00:06<00:00, 3.83it/s] 93%|█████████▎| 27/29 [00:07<00:00, 3.82it/s] 97%|█████████▋| 28/29 [00:07<00:00, 3.83it/s] 100%|██████████| 29/29 [00:07<00:00, 3.83it/s] 100%|██████████| 29/29 [00:07<00:00, 3.84it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID3nyh1zhhdxrmc0cp9w1ah1pc48StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc Style: regal ethereal fantasy Composition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form Lighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs Color Palette: regal purples, soft golds, pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace in motion Technical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns Additional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.21
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 50
{ "model": "dev", "prompt": "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.21, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 50 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.21' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=50'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T17:04:39.893228Z", "created_at": "2025-04-19T17:04:25.967000Z", "data_removed": false, "error": null, "id": "3nyh1zhhdxrmc0cp9w1ah1pc48", "input": { "model": "dev", "prompt": "Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 29836\nPrompt: Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc\nStyle: regal ethereal fantasy\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue\nMood/Atmosphere: majestic calm, otherworldly grace in motion\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/50 [00:00<?, ?it/s]\n 2%|▏ | 1/50 [00:00<00:13, 3.75it/s]\n 4%|▍ | 2/50 [00:00<00:11, 4.16it/s]\n 6%|▌ | 3/50 [00:00<00:11, 4.01it/s]\n 8%|▊ | 4/50 [00:01<00:11, 3.94it/s]\n 10%|█ | 5/50 [00:01<00:11, 3.91it/s]\n 12%|█▏ | 6/50 [00:01<00:11, 3.88it/s]\n 14%|█▍ | 7/50 [00:01<00:11, 3.87it/s]\n 16%|█▌ | 8/50 [00:02<00:10, 3.86it/s]\n 18%|█▊ | 9/50 [00:02<00:10, 3.85it/s]\n 20%|██ | 10/50 [00:02<00:10, 3.85it/s]\n 22%|██▏ | 11/50 [00:02<00:10, 3.85it/s]\n 24%|██▍ | 12/50 [00:03<00:09, 3.85it/s]\n 26%|██▌ | 13/50 [00:03<00:09, 3.84it/s]\n 28%|██▊ | 14/50 [00:03<00:09, 3.84it/s]\n 30%|███ | 15/50 [00:03<00:09, 3.84it/s]\n 32%|███▏ | 16/50 [00:04<00:08, 3.85it/s]\n 34%|███▍ | 17/50 [00:04<00:08, 3.84it/s]\n 36%|███▌ | 18/50 [00:04<00:08, 3.84it/s]\n 38%|███▊ | 19/50 [00:04<00:08, 3.83it/s]\n 40%|████ | 20/50 [00:05<00:07, 3.84it/s]\n 42%|████▏ | 21/50 [00:05<00:07, 3.82it/s]\n 44%|████▍ | 22/50 [00:05<00:07, 3.84it/s]\n 46%|████▌ | 23/50 [00:05<00:07, 3.83it/s]\n 48%|████▊ | 24/50 [00:06<00:06, 3.84it/s]\n 50%|█████ | 25/50 [00:06<00:06, 3.83it/s]\n 52%|█████▏ | 26/50 [00:06<00:06, 3.83it/s]\n 54%|█████▍ | 27/50 [00:07<00:06, 3.83it/s]\n 56%|█████▌ | 28/50 [00:07<00:05, 3.84it/s]\n 58%|█████▊ | 29/50 [00:07<00:05, 3.84it/s]\n 60%|██████ | 30/50 [00:07<00:05, 3.84it/s]\n 62%|██████▏ | 31/50 [00:08<00:04, 3.84it/s]\n 64%|██████▍ | 32/50 [00:08<00:04, 3.84it/s]\n 66%|██████▌ | 33/50 [00:08<00:04, 3.84it/s]\n 68%|██████▊ | 34/50 [00:08<00:04, 3.84it/s]\n 70%|███████ | 35/50 [00:09<00:03, 3.84it/s]\n 72%|███████▏ | 36/50 [00:09<00:03, 3.83it/s]\n 74%|███████▍ | 37/50 [00:09<00:03, 3.83it/s]\n 76%|███████▌ | 38/50 [00:09<00:03, 3.83it/s]\n 78%|███████▊ | 39/50 [00:10<00:02, 3.83it/s]\n 80%|████████ | 40/50 [00:10<00:02, 3.84it/s]\n 82%|████████▏ | 41/50 [00:10<00:02, 3.85it/s]\n 84%|████████▍ | 42/50 [00:10<00:02, 3.84it/s]\n 86%|████████▌ | 43/50 [00:11<00:01, 3.85it/s]\n 88%|████████▊ | 44/50 [00:11<00:01, 3.84it/s]\n 90%|█████████ | 45/50 [00:11<00:01, 3.85it/s]\n 92%|█████████▏| 46/50 [00:11<00:01, 3.84it/s]\n 94%|█████████▍| 47/50 [00:12<00:00, 3.85it/s]\n 96%|█████████▌| 48/50 [00:12<00:00, 3.84it/s]\n 98%|█████████▊| 49/50 [00:12<00:00, 3.84it/s]\n100%|██████████| 50/50 [00:12<00:00, 3.84it/s]\n100%|██████████| 50/50 [00:12<00:00, 3.85it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 13.847551998, "total_time": 13.926228 }, "output": [ "https://replicate.delivery/xezq/ru4BSem4a2yON6lbdskA7Mzy4SR0S9wHlcxi8Qv3hP2TTMSKA/out-0.webp" ], "started_at": "2025-04-19T17:04:26.045676Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-oxs2mpe4fw2nxmy5bem27m2p2rebqwjm5nb7fdut2k6gc63mvc7q", "get": "https://api.replicate.com/v1/predictions/3nyh1zhhdxrmc0cp9w1ah1pc48", "cancel": "https://api.replicate.com/v1/predictions/3nyh1zhhdxrmc0cp9w1ah1pc48/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 29836 Prompt: Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc Style: regal ethereal fantasy Composition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form Lighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs Color Palette: regal purples, soft golds, pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace in motion Technical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns Additional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/50 [00:00<?, ?it/s] 2%|▏ | 1/50 [00:00<00:13, 3.75it/s] 4%|▍ | 2/50 [00:00<00:11, 4.16it/s] 6%|▌ | 3/50 [00:00<00:11, 4.01it/s] 8%|▊ | 4/50 [00:01<00:11, 3.94it/s] 10%|█ | 5/50 [00:01<00:11, 3.91it/s] 12%|█▏ | 6/50 [00:01<00:11, 3.88it/s] 14%|█▍ | 7/50 [00:01<00:11, 3.87it/s] 16%|█▌ | 8/50 [00:02<00:10, 3.86it/s] 18%|█▊ | 9/50 [00:02<00:10, 3.85it/s] 20%|██ | 10/50 [00:02<00:10, 3.85it/s] 22%|██▏ | 11/50 [00:02<00:10, 3.85it/s] 24%|██▍ | 12/50 [00:03<00:09, 3.85it/s] 26%|██▌ | 13/50 [00:03<00:09, 3.84it/s] 28%|██▊ | 14/50 [00:03<00:09, 3.84it/s] 30%|███ | 15/50 [00:03<00:09, 3.84it/s] 32%|███▏ | 16/50 [00:04<00:08, 3.85it/s] 34%|███▍ | 17/50 [00:04<00:08, 3.84it/s] 36%|███▌ | 18/50 [00:04<00:08, 3.84it/s] 38%|███▊ | 19/50 [00:04<00:08, 3.83it/s] 40%|████ | 20/50 [00:05<00:07, 3.84it/s] 42%|████▏ | 21/50 [00:05<00:07, 3.82it/s] 44%|████▍ | 22/50 [00:05<00:07, 3.84it/s] 46%|████▌ | 23/50 [00:05<00:07, 3.83it/s] 48%|████▊ | 24/50 [00:06<00:06, 3.84it/s] 50%|█████ | 25/50 [00:06<00:06, 3.83it/s] 52%|█████▏ | 26/50 [00:06<00:06, 3.83it/s] 54%|█████▍ | 27/50 [00:07<00:06, 3.83it/s] 56%|█████▌ | 28/50 [00:07<00:05, 3.84it/s] 58%|█████▊ | 29/50 [00:07<00:05, 3.84it/s] 60%|██████ | 30/50 [00:07<00:05, 3.84it/s] 62%|██████▏ | 31/50 [00:08<00:04, 3.84it/s] 64%|██████▍ | 32/50 [00:08<00:04, 3.84it/s] 66%|██████▌ | 33/50 [00:08<00:04, 3.84it/s] 68%|██████▊ | 34/50 [00:08<00:04, 3.84it/s] 70%|███████ | 35/50 [00:09<00:03, 3.84it/s] 72%|███████▏ | 36/50 [00:09<00:03, 3.83it/s] 74%|███████▍ | 37/50 [00:09<00:03, 3.83it/s] 76%|███████▌ | 38/50 [00:09<00:03, 3.83it/s] 78%|███████▊ | 39/50 [00:10<00:02, 3.83it/s] 80%|████████ | 40/50 [00:10<00:02, 3.84it/s] 82%|████████▏ | 41/50 [00:10<00:02, 3.85it/s] 84%|████████▍ | 42/50 [00:10<00:02, 3.84it/s] 86%|████████▌ | 43/50 [00:11<00:01, 3.85it/s] 88%|████████▊ | 44/50 [00:11<00:01, 3.84it/s] 90%|█████████ | 45/50 [00:11<00:01, 3.85it/s] 92%|█████████▏| 46/50 [00:11<00:01, 3.84it/s] 94%|█████████▍| 47/50 [00:12<00:00, 3.85it/s] 96%|█████████▌| 48/50 [00:12<00:00, 3.84it/s] 98%|█████████▊| 49/50 [00:12<00:00, 3.84it/s] 100%|██████████| 50/50 [00:12<00:00, 3.84it/s] 100%|██████████| 50/50 [00:12<00:00, 3.85it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDt9hsfsp641rme0cp9w2va42jtrStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc Style: regal ethereal fantasy Composition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form Lighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs Color Palette: regal purples, soft golds, pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace in motion Technical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns Additional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.21
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 50
{ "model": "dev", "prompt": " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.21, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 50 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt=" Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.21' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=50'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \\nStyle: regal ethereal fantasy \\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \\nMood/Atmosphere: majestic calm, otherworldly grace in motion \\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T17:08:37.158646Z", "created_at": "2025-04-19T17:08:20.640000Z", "data_removed": false, "error": null, "id": "t9hsfsp641rme0cp9w2va42jtr", "input": { "model": "dev", "prompt": " Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc \nStyle: regal ethereal fantasy \nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form \nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs \nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue \nMood/Atmosphere: majestic calm, otherworldly grace in motion \nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns \nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose \n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.21, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 50 }, "logs": "free=25383657910272\nDownloading weights\n2025-04-19T17:08:21Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpg379qa27/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T17:08:21Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T17:08:23Z | INFO | [ Complete ] dest=/tmp/tmpg379qa27/weights size=\"172 MB\" total_elapsed=1.893s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 1.91s\nLoaded LoRAs in 2.49s\nUsing seed: 26085\nPrompt: Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc\nStyle: regal ethereal fantasy\nComposition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form\nLighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs\nColor Palette: regal purples, soft golds, pearlescent whites with hints of opal blue\nMood/Atmosphere: majestic calm, otherworldly grace in motion\nTechnical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns\nAdditional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose\n--ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55\n[!] txt2img mode\n 0%| | 0/50 [00:00<?, ?it/s]\n 2%|▏ | 1/50 [00:00<00:13, 3.74it/s]\n 4%|▍ | 2/50 [00:00<00:11, 4.16it/s]\n 6%|▌ | 3/50 [00:00<00:11, 4.00it/s]\n 8%|▊ | 4/50 [00:01<00:11, 3.91it/s]\n 10%|█ | 5/50 [00:01<00:11, 3.88it/s]\n 12%|█▏ | 6/50 [00:01<00:11, 3.87it/s]\n 14%|█▍ | 7/50 [00:01<00:11, 3.84it/s]\n 16%|█▌ | 8/50 [00:02<00:10, 3.83it/s]\n 18%|█▊ | 9/50 [00:02<00:10, 3.83it/s]\n 20%|██ | 10/50 [00:02<00:10, 3.83it/s]\n 22%|██▏ | 11/50 [00:02<00:10, 3.82it/s]\n 24%|██▍ | 12/50 [00:03<00:09, 3.82it/s]\n 26%|██▌ | 13/50 [00:03<00:09, 3.82it/s]\n 28%|██▊ | 14/50 [00:03<00:09, 3.83it/s]\n 30%|███ | 15/50 [00:03<00:09, 3.82it/s]\n 32%|███▏ | 16/50 [00:04<00:08, 3.82it/s]\n 34%|███▍ | 17/50 [00:04<00:08, 3.82it/s]\n 36%|███▌ | 18/50 [00:04<00:08, 3.82it/s]\n 38%|███▊ | 19/50 [00:04<00:08, 3.81it/s]\n 40%|████ | 20/50 [00:05<00:07, 3.82it/s]\n 42%|████▏ | 21/50 [00:05<00:07, 3.83it/s]\n 44%|████▍ | 22/50 [00:05<00:07, 3.82it/s]\n 46%|████▌ | 23/50 [00:05<00:07, 3.81it/s]\n 48%|████▊ | 24/50 [00:06<00:06, 3.82it/s]\n 50%|█████ | 25/50 [00:06<00:06, 3.82it/s]\n 52%|█████▏ | 26/50 [00:06<00:06, 3.82it/s]\n 54%|█████▍ | 27/50 [00:07<00:06, 3.81it/s]\n 56%|█████▌ | 28/50 [00:07<00:05, 3.82it/s]\n 58%|█████▊ | 29/50 [00:07<00:05, 3.82it/s]\n 60%|██████ | 30/50 [00:07<00:05, 3.81it/s]\n 62%|██████▏ | 31/50 [00:08<00:04, 3.82it/s]\n 64%|██████▍ | 32/50 [00:08<00:04, 3.81it/s]\n 66%|██████▌ | 33/50 [00:08<00:04, 3.82it/s]\n 68%|██████▊ | 34/50 [00:08<00:04, 3.81it/s]\n 70%|███████ | 35/50 [00:09<00:03, 3.81it/s]\n 72%|███████▏ | 36/50 [00:09<00:03, 3.82it/s]\n 74%|███████▍ | 37/50 [00:09<00:03, 3.82it/s]\n 76%|███████▌ | 38/50 [00:09<00:03, 3.81it/s]\n 78%|███████▊ | 39/50 [00:10<00:02, 3.81it/s]\n 80%|████████ | 40/50 [00:10<00:02, 3.82it/s]\n 82%|████████▏ | 41/50 [00:10<00:02, 3.81it/s]\n 84%|████████▍ | 42/50 [00:10<00:02, 3.81it/s]\n 86%|████████▌ | 43/50 [00:11<00:01, 3.82it/s]\n 88%|████████▊ | 44/50 [00:11<00:01, 3.82it/s]\n 90%|█████████ | 45/50 [00:11<00:01, 3.81it/s]\n 92%|█████████▏| 46/50 [00:12<00:01, 3.80it/s]\n 94%|█████████▍| 47/50 [00:12<00:00, 3.81it/s]\n 96%|█████████▌| 48/50 [00:12<00:00, 3.81it/s]\n 98%|█████████▊| 49/50 [00:12<00:00, 3.81it/s]\n100%|██████████| 50/50 [00:13<00:00, 3.81it/s]\n100%|██████████| 50/50 [00:13<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 15.849829578, "total_time": 16.518646 }, "output": [ "https://replicate.delivery/xezq/LjPDR1cKI0ZjMdWfcXeAa5F7WWRmbCAqJrwZdztvzkHVqYkUA/out-0.webp" ], "started_at": "2025-04-19T17:08:21.308817Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-2sv2alsjwhibshp26qrz6mjdgn2wdwttnjyaye5hpk3zxbs5hryq", "get": "https://api.replicate.com/v1/predictions/t9hsfsp641rme0cp9w2va42jtr", "cancel": "https://api.replicate.com/v1/predictions/t9hsfsp641rme0cp9w2va42jtr/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25383657910272 Downloading weights 2025-04-19T17:08:21Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmpg379qa27/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T17:08:21Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T17:08:23Z | INFO | [ Complete ] dest=/tmp/tmpg379qa27/weights size="172 MB" total_elapsed=1.893s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 1.91s Loaded LoRAs in 2.49s Using seed: 26085 Prompt: Subject: TOKMAHI in a dynamic, soaring pose—her full face and neck visible as she reclines on a tilted marble shard, one arm stretched above her head, the other bent gracefully across her chest, legs curled in a ballet‑like arc Style: regal ethereal fantasy Composition: overhead, tight square crop from shoulders to mid‑thigh, inverted‑gravity clouds and drifting lanterns spiraling around her form Lighting: soft divine beams from above casting a warm golden aura on her skin and hands, crisp rim highlights tracing the silhouette of her extended limbs Color Palette: regal purples, soft golds, pearlescent whites with hints of opal blue Mood/Atmosphere: majestic calm, otherworldly grace in motion Technical Details: tack‑sharp focus on her eyes, face, and hands; gentle motion blur on cloud wisps and trailing lanterns Additional Elements: faint constellations forming a diadem above her head, shimmering motes of light rising like incense smoke around her pose --ar 1:1 --v flux1 --q 2 --style dreamlike --chaos 55 [!] txt2img mode 0%| | 0/50 [00:00<?, ?it/s] 2%|▏ | 1/50 [00:00<00:13, 3.74it/s] 4%|▍ | 2/50 [00:00<00:11, 4.16it/s] 6%|▌ | 3/50 [00:00<00:11, 4.00it/s] 8%|▊ | 4/50 [00:01<00:11, 3.91it/s] 10%|█ | 5/50 [00:01<00:11, 3.88it/s] 12%|█▏ | 6/50 [00:01<00:11, 3.87it/s] 14%|█▍ | 7/50 [00:01<00:11, 3.84it/s] 16%|█▌ | 8/50 [00:02<00:10, 3.83it/s] 18%|█▊ | 9/50 [00:02<00:10, 3.83it/s] 20%|██ | 10/50 [00:02<00:10, 3.83it/s] 22%|██▏ | 11/50 [00:02<00:10, 3.82it/s] 24%|██▍ | 12/50 [00:03<00:09, 3.82it/s] 26%|██▌ | 13/50 [00:03<00:09, 3.82it/s] 28%|██▊ | 14/50 [00:03<00:09, 3.83it/s] 30%|███ | 15/50 [00:03<00:09, 3.82it/s] 32%|███▏ | 16/50 [00:04<00:08, 3.82it/s] 34%|███▍ | 17/50 [00:04<00:08, 3.82it/s] 36%|███▌ | 18/50 [00:04<00:08, 3.82it/s] 38%|███▊ | 19/50 [00:04<00:08, 3.81it/s] 40%|████ | 20/50 [00:05<00:07, 3.82it/s] 42%|████▏ | 21/50 [00:05<00:07, 3.83it/s] 44%|████▍ | 22/50 [00:05<00:07, 3.82it/s] 46%|████▌ | 23/50 [00:05<00:07, 3.81it/s] 48%|████▊ | 24/50 [00:06<00:06, 3.82it/s] 50%|█████ | 25/50 [00:06<00:06, 3.82it/s] 52%|█████▏ | 26/50 [00:06<00:06, 3.82it/s] 54%|█████▍ | 27/50 [00:07<00:06, 3.81it/s] 56%|█████▌ | 28/50 [00:07<00:05, 3.82it/s] 58%|█████▊ | 29/50 [00:07<00:05, 3.82it/s] 60%|██████ | 30/50 [00:07<00:05, 3.81it/s] 62%|██████▏ | 31/50 [00:08<00:04, 3.82it/s] 64%|██████▍ | 32/50 [00:08<00:04, 3.81it/s] 66%|██████▌ | 33/50 [00:08<00:04, 3.82it/s] 68%|██████▊ | 34/50 [00:08<00:04, 3.81it/s] 70%|███████ | 35/50 [00:09<00:03, 3.81it/s] 72%|███████▏ | 36/50 [00:09<00:03, 3.82it/s] 74%|███████▍ | 37/50 [00:09<00:03, 3.82it/s] 76%|███████▌ | 38/50 [00:09<00:03, 3.81it/s] 78%|███████▊ | 39/50 [00:10<00:02, 3.81it/s] 80%|████████ | 40/50 [00:10<00:02, 3.82it/s] 82%|████████▏ | 41/50 [00:10<00:02, 3.81it/s] 84%|████████▍ | 42/50 [00:10<00:02, 3.81it/s] 86%|████████▌ | 43/50 [00:11<00:01, 3.82it/s] 88%|████████▊ | 44/50 [00:11<00:01, 3.82it/s] 90%|█████████ | 45/50 [00:11<00:01, 3.81it/s] 92%|█████████▏| 46/50 [00:12<00:01, 3.80it/s] 94%|█████████▍| 47/50 [00:12<00:00, 3.81it/s] 96%|█████████▌| 48/50 [00:12<00:00, 3.81it/s] 98%|█████████▊| 49/50 [00:12<00:00, 3.81it/s] 100%|██████████| 50/50 [00:13<00:00, 3.81it/s] 100%|██████████| 50/50 [00:13<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID6dcnfrrdd1rme0cp9ww84hsb7gStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer Style: avant‑garde editorial fusion Composition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline Lighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows Color Palette: electric magentas, deep indigos, molten silvers Mood/Atmosphere: surreal tension, boundary‑pushing chaos Technical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips Additional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette --ar 1:1 --v flux1 --q 2 --style raw --chaos 80
- go_fast
- lora_scale
- 1.043
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.38
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \nStyle: avant‑garde editorial fusion \nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \nColor Palette: electric magentas, deep indigos, molten silvers \nMood/Atmosphere: surreal tension, boundary‑pushing chaos \nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.38, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \nStyle: avant‑garde editorial fusion \nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \nColor Palette: electric magentas, deep indigos, molten silvers \nMood/Atmosphere: surreal tension, boundary‑pushing chaos \nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\n", go_fast: false, lora_scale: 1.043, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.38, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \nStyle: avant‑garde editorial fusion \nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \nColor Palette: electric magentas, deep indigos, molten silvers \nMood/Atmosphere: surreal tension, boundary‑pushing chaos \nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": False, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.38, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \\nStyle: avant‑garde editorial fusion \\nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \\nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \\nColor Palette: electric magentas, deep indigos, molten silvers \\nMood/Atmosphere: surreal tension, boundary‑pushing chaos \\nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \\nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.38, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \\nStyle: avant‑garde editorial fusion \\nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \\nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \\nColor Palette: electric magentas, deep indigos, molten silvers \\nMood/Atmosphere: surreal tension, boundary‑pushing chaos \\nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \\nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.043' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.38' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \\nStyle: avant‑garde editorial fusion \\nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \\nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \\nColor Palette: electric magentas, deep indigos, molten silvers \\nMood/Atmosphere: surreal tension, boundary‑pushing chaos \\nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \\nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.38, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:03:25.071854Z", "created_at": "2025-04-19T18:03:15.688000Z", "data_removed": false, "error": null, "id": "6dcnfrrdd1rme0cp9ww84hsb7g", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer \nStyle: avant‑garde editorial fusion \nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline \nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows \nColor Palette: electric magentas, deep indigos, molten silvers \nMood/Atmosphere: surreal tension, boundary‑pushing chaos \nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips \nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.043, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.38, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "free=24854523674624\nDownloading weights\n2025-04-19T18:03:15Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp080o6fpv/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T18:03:15Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T18:03:16Z | INFO | [ Complete ] dest=/tmp/tmp080o6fpv/weights size=\"172 MB\" total_elapsed=0.594s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 0.61s\nLoaded LoRAs in 1.19s\nUsing seed: 60343\nPrompt: Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer\nStyle: avant‑garde editorial fusion\nComposition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline\nLighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows\nColor Palette: electric magentas, deep indigos, molten silvers\nMood/Atmosphere: surreal tension, boundary‑pushing chaos\nTechnical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips\nAdditional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 80\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.71it/s]\n 7%|▋ | 2/30 [00:00<00:07, 3.79it/s]\n 10%|█ | 3/30 [00:00<00:07, 3.84it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.83it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.81it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.83it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.84it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.85it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.83it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.86it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.84it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.84it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.84it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.83it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.83it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.83it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.83it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s]\n 77%|███████▋ | 23/30 [00:06<00:01, 3.83it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.79it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.80it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.82it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.80it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 9.337994804, "total_time": 9.383854 }, "output": [ "https://replicate.delivery/xezq/b0uc4vVTDZpRCFU4dRdo9GAWnve72ZITWve58hIxm2rtdZkUA/out-0.webp" ], "started_at": "2025-04-19T18:03:15.733859Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-yrqmbg2gwyg2z6jim3fm5vzbox6k3eow62qnxkbfdu64oyed3ewa", "get": "https://api.replicate.com/v1/predictions/6dcnfrrdd1rme0cp9ww84hsb7g", "cancel": "https://api.replicate.com/v1/predictions/6dcnfrrdd1rme0cp9ww84hsb7g/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=24854523674624 Downloading weights 2025-04-19T18:03:15Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp080o6fpv/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T18:03:15Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T18:03:16Z | INFO | [ Complete ] dest=/tmp/tmp080o6fpv/weights size="172 MB" total_elapsed=0.594s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 0.61s Loaded LoRAs in 1.19s Using seed: 60343 Prompt: Subject: TOKMAHI’s full face and neck in tight portrait, eyes meeting the viewer Style: avant‑garde editorial fusion Composition: inverted perspective, tilted axis, extreme close‑up from shoulders to hairline Lighting: harsh rim light slicing through drifting mist, punctuated by soft bioluminescent glows Color Palette: electric magentas, deep indigos, molten silvers Mood/Atmosphere: surreal tension, boundary‑pushing chaos Technical Details: specify Sigma 75 mm lens, f/1.4 depth of field, pronounced motion blur on hair tips Additional Elements: floating metallic shards, swirling fractal ribbons, gravity distortions around her silhouette --ar 1:1 --v flux1 --q 2 --style raw --chaos 80 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.71it/s] 7%|▋ | 2/30 [00:00<00:07, 3.79it/s] 10%|█ | 3/30 [00:00<00:07, 3.84it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.83it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.81it/s] 20%|██ | 6/30 [00:01<00:06, 3.83it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.84it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.85it/s] 30%|███ | 9/30 [00:02<00:05, 3.83it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.86it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.85it/s] 40%|████ | 12/30 [00:03<00:04, 3.84it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.84it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.84it/s] 50%|█████ | 15/30 [00:03<00:03, 3.84it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.84it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.83it/s] 60%|██████ | 18/30 [00:04<00:03, 3.83it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.83it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.83it/s] 70%|███████ | 21/30 [00:05<00:02, 3.83it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.83it/s] 77%|███████▋ | 23/30 [00:06<00:01, 3.83it/s] 80%|████████ | 24/30 [00:06<00:01, 3.79it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.80it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.82it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.82it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.80it/s] 100%|██████████| 30/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDf5rk9bm1esrma0cp9x099ygew4StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85
- go_fast
- lora_scale
- 1.019
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.2
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 31
{ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.2, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 31 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", go_fast: false, lora_scale: 1.019, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.2, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 31 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": False, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.2, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 31 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.2, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 31 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.019' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.2' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=31'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \\nStyle: avant‑garde deconstructive abstraction \\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \\nMood/Atmosphere: dissonant elegance, fractured surrealism \\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.2, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 31 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:12:47.774775Z", "created_at": "2025-04-19T18:12:29.686000Z", "data_removed": false, "error": null, "id": "f5rk9bm1esrma0cp9x099ygew4", "input": { "model": "dev", "prompt": "Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction \nStyle: avant‑garde deconstructive abstraction \nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes \nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles \nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents \nMood/Atmosphere: dissonant elegance, fractured surrealism \nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges \nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n", "go_fast": false, "lora_scale": 1.019, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.2, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 31 }, "logs": "free=25597805617152\nDownloading weights\n2025-04-19T18:12:29Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp4jmlabt1/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\n2025-04-19T18:12:29Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local\n2025-04-19T18:12:36Z | INFO | [ Complete ] dest=/tmp/tmp4jmlabt1/weights size=\"172 MB\" total_elapsed=7.112s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar\nDownloaded weights in 7.13s\nLoaded LoRAs in 9.65s\nUsing seed: 36461\nPrompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction\nStyle: avant‑garde deconstructive abstraction\nComposition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes\nLighting: sharp metallic rim light intercut with diffuse neon glows from random angles\nColor Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents\nMood/Atmosphere: dissonant elegance, fractured surrealism\nTechnical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges\nAdditional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 85\n[!] txt2img mode\n 0%| | 0/31 [00:00<?, ?it/s]\n 3%|▎ | 1/31 [00:00<00:08, 3.72it/s]\n 6%|▋ | 2/31 [00:00<00:06, 4.15it/s]\n 10%|▉ | 3/31 [00:00<00:07, 3.98it/s]\n 13%|█▎ | 4/31 [00:01<00:06, 3.92it/s]\n 16%|█▌ | 5/31 [00:01<00:06, 3.89it/s]\n 19%|█▉ | 6/31 [00:01<00:06, 3.85it/s]\n 23%|██▎ | 7/31 [00:01<00:06, 3.84it/s]\n 26%|██▌ | 8/31 [00:02<00:05, 3.84it/s]\n 29%|██▉ | 9/31 [00:02<00:05, 3.83it/s]\n 32%|███▏ | 10/31 [00:02<00:05, 3.82it/s]\n 35%|███▌ | 11/31 [00:02<00:05, 3.82it/s]\n 39%|███▊ | 12/31 [00:03<00:04, 3.82it/s]\n 42%|████▏ | 13/31 [00:03<00:04, 3.82it/s]\n 45%|████▌ | 14/31 [00:03<00:04, 3.82it/s]\n 48%|████▊ | 15/31 [00:03<00:04, 3.82it/s]\n 52%|█████▏ | 16/31 [00:04<00:03, 3.82it/s]\n 55%|█████▍ | 17/31 [00:04<00:03, 3.82it/s]\n 58%|█████▊ | 18/31 [00:04<00:03, 3.79it/s]\n 61%|██████▏ | 19/31 [00:04<00:03, 3.82it/s]\n 65%|██████▍ | 20/31 [00:05<00:02, 3.82it/s]\n 68%|██████▊ | 21/31 [00:05<00:02, 3.82it/s]\n 71%|███████ | 22/31 [00:05<00:02, 3.82it/s]\n 74%|███████▍ | 23/31 [00:05<00:02, 3.82it/s]\n 77%|███████▋ | 24/31 [00:06<00:01, 3.82it/s]\n 81%|████████ | 25/31 [00:06<00:01, 3.81it/s]\n 84%|████████▍ | 26/31 [00:06<00:01, 3.82it/s]\n 87%|████████▋ | 27/31 [00:07<00:01, 3.82it/s]\n 90%|█████████ | 28/31 [00:07<00:00, 3.82it/s]\n 94%|█████████▎| 29/31 [00:07<00:00, 3.82it/s]\n 97%|█████████▋| 30/31 [00:07<00:00, 3.82it/s]\n100%|██████████| 31/31 [00:08<00:00, 3.82it/s]\n100%|██████████| 31/31 [00:08<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 18.024718957, "total_time": 18.088775 }, "output": [ "https://replicate.delivery/xezq/GBnz0kVcOro5OFIfagQ58J9OzMgOcvjNKUoxHXfms3CfMzIpA/out-0.webp" ], "started_at": "2025-04-19T18:12:29.750056Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-jhqzwoxdxs5rh4vis23mnyfq5v7oovlias27dhibximzsswuhvba", "get": "https://api.replicate.com/v1/predictions/f5rk9bm1esrma0cp9x099ygew4", "cancel": "https://api.replicate.com/v1/predictions/f5rk9bm1esrma0cp9x099ygew4/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated infree=25597805617152 Downloading weights 2025-04-19T18:12:29Z | INFO | [ Initiating ] chunk_size=150M dest=/tmp/tmp4jmlabt1/weights url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar 2025-04-19T18:12:29Z | INFO | [ Cache Service ] enabled=true scheme=http target=hermes.services.svc.cluster.local 2025-04-19T18:12:36Z | INFO | [ Complete ] dest=/tmp/tmp4jmlabt1/weights size="172 MB" total_elapsed=7.112s url=https://replicate.delivery/xezq/pnQLbBjzqK79NBDhefLNajGzBl0yOqxY0UHn8QTFQlomLEkUA/trained_model.tar Downloaded weights in 7.13s Loaded LoRAs in 9.65s Using seed: 36461 Prompt: Subject: TOKMAHI’s full face and neck in extreme close‑up, eyes piercing through layered abstraction Style: avant‑garde deconstructive abstraction Composition: double‑exposure effect merging her portrait with jagged geometric glass shards, tilted 45° axis, fragmented planes Lighting: sharp metallic rim light intercut with diffuse neon glows from random angles Color Palette: stark blacks and whites punctuated by neon cyan, hot magenta, and acid yellow accents Mood/Atmosphere: dissonant elegance, fractured surrealism Technical Details: use a 50 mm tilt‑shift lens simulation, f/2.0 depth of field, heavy motion blur on shard edges Additional Elements: glitch‑style artifacts, swirling smoke ribbons, reflective metal flakes suspended around her --ar 1:1 --v flux1 --q 2 --style raw --chaos 85 [!] txt2img mode 0%| | 0/31 [00:00<?, ?it/s] 3%|▎ | 1/31 [00:00<00:08, 3.72it/s] 6%|▋ | 2/31 [00:00<00:06, 4.15it/s] 10%|▉ | 3/31 [00:00<00:07, 3.98it/s] 13%|█▎ | 4/31 [00:01<00:06, 3.92it/s] 16%|█▌ | 5/31 [00:01<00:06, 3.89it/s] 19%|█▉ | 6/31 [00:01<00:06, 3.85it/s] 23%|██▎ | 7/31 [00:01<00:06, 3.84it/s] 26%|██▌ | 8/31 [00:02<00:05, 3.84it/s] 29%|██▉ | 9/31 [00:02<00:05, 3.83it/s] 32%|███▏ | 10/31 [00:02<00:05, 3.82it/s] 35%|███▌ | 11/31 [00:02<00:05, 3.82it/s] 39%|███▊ | 12/31 [00:03<00:04, 3.82it/s] 42%|████▏ | 13/31 [00:03<00:04, 3.82it/s] 45%|████▌ | 14/31 [00:03<00:04, 3.82it/s] 48%|████▊ | 15/31 [00:03<00:04, 3.82it/s] 52%|█████▏ | 16/31 [00:04<00:03, 3.82it/s] 55%|█████▍ | 17/31 [00:04<00:03, 3.82it/s] 58%|█████▊ | 18/31 [00:04<00:03, 3.79it/s] 61%|██████▏ | 19/31 [00:04<00:03, 3.82it/s] 65%|██████▍ | 20/31 [00:05<00:02, 3.82it/s] 68%|██████▊ | 21/31 [00:05<00:02, 3.82it/s] 71%|███████ | 22/31 [00:05<00:02, 3.82it/s] 74%|███████▍ | 23/31 [00:05<00:02, 3.82it/s] 77%|███████▋ | 24/31 [00:06<00:01, 3.82it/s] 81%|████████ | 25/31 [00:06<00:01, 3.81it/s] 84%|████████▍ | 26/31 [00:06<00:01, 3.82it/s] 87%|████████▋ | 27/31 [00:07<00:01, 3.82it/s] 90%|█████████ | 28/31 [00:07<00:00, 3.82it/s] 94%|█████████▎| 29/31 [00:07<00:00, 3.82it/s] 97%|█████████▋| 30/31 [00:07<00:00, 3.82it/s] 100%|██████████| 31/31 [00:08<00:00, 3.82it/s] 100%|██████████| 31/31 [00:08<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104ID3qpexhp4v1rm80cp9x5srjfrvcStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines Style: avant-garde deconstructive elegance Composition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth Lighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows Color Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones Mood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit Technical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds Additional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris --ar 3:4 --v flux1 --q 2 --style raw --chaos 80
- go_fast
- lora_scale
- 1.104
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.3
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 36
{ "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 36 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", go_fast: false, lora_scale: 1.104, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.3, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 36 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": False, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 36 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 36 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.104' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.3' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=36'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women\'s jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \\nStyle: avant-garde deconstructive elegance \\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 36 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T18:24:58.135890Z", "created_at": "2025-04-19T18:24:47.832000Z", "data_removed": false, "error": null, "id": "3qpexhp4v1rm80cp9x5srjfrvc", "input": { "model": "dev", "prompt": "Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines \nStyle: avant-garde deconstructive elegance \nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth \nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows \nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones \nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit \nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds \nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris \n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n", "go_fast": false, "lora_scale": 1.104, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.3, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 36 }, "logs": "Loaded LoRAs in 0.58s\nUsing seed: 9412\nPrompt: Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines\nStyle: avant-garde deconstructive elegance\nComposition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth\nLighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows\nColor Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones\nMood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit\nTechnical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds\nAdditional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris\n--ar 3:4 --v flux1 --q 2 --style raw --chaos 80\n[!] txt2img mode\n 0%| | 0/36 [00:00<?, ?it/s]\n 3%|▎ | 1/36 [00:00<00:09, 3.69it/s]\n 6%|▌ | 2/36 [00:00<00:08, 4.08it/s]\n 8%|▊ | 3/36 [00:00<00:08, 3.95it/s]\n 11%|█ | 4/36 [00:01<00:08, 3.90it/s]\n 14%|█▍ | 5/36 [00:01<00:08, 3.87it/s]\n 17%|█▋ | 6/36 [00:01<00:07, 3.85it/s]\n 19%|█▉ | 7/36 [00:01<00:07, 3.84it/s]\n 22%|██▏ | 8/36 [00:02<00:07, 3.83it/s]\n 25%|██▌ | 9/36 [00:02<00:07, 3.83it/s]\n 28%|██▊ | 10/36 [00:02<00:06, 3.83it/s]\n 31%|███ | 11/36 [00:02<00:06, 3.82it/s]\n 33%|███▎ | 12/36 [00:03<00:06, 3.82it/s]\n 36%|███▌ | 13/36 [00:03<00:06, 3.83it/s]\n 39%|███▉ | 14/36 [00:03<00:05, 3.82it/s]\n 42%|████▏ | 15/36 [00:03<00:05, 3.81it/s]\n 44%|████▍ | 16/36 [00:04<00:05, 3.82it/s]\n 47%|████▋ | 17/36 [00:04<00:04, 3.82it/s]\n 50%|█████ | 18/36 [00:04<00:04, 3.80it/s]\n 53%|█████▎ | 19/36 [00:04<00:04, 3.81it/s]\n 56%|█████▌ | 20/36 [00:05<00:04, 3.81it/s]\n 58%|█████▊ | 21/36 [00:05<00:03, 3.81it/s]\n 61%|██████ | 22/36 [00:05<00:03, 3.81it/s]\n 64%|██████▍ | 23/36 [00:06<00:03, 3.81it/s]\n 67%|██████▋ | 24/36 [00:06<00:03, 3.81it/s]\n 69%|██████▉ | 25/36 [00:06<00:02, 3.81it/s]\n 72%|███████▏ | 26/36 [00:06<00:02, 3.80it/s]\n 75%|███████▌ | 27/36 [00:07<00:02, 3.80it/s]\n 78%|███████▊ | 28/36 [00:07<00:02, 3.81it/s]\n 81%|████████ | 29/36 [00:07<00:01, 3.81it/s]\n 83%|████████▎ | 30/36 [00:07<00:01, 3.80it/s]\n 86%|████████▌ | 31/36 [00:08<00:01, 3.80it/s]\n 89%|████████▉ | 32/36 [00:08<00:01, 3.81it/s]\n 92%|█████████▏| 33/36 [00:08<00:00, 3.81it/s]\n 94%|█████████▍| 34/36 [00:08<00:00, 3.80it/s]\n 97%|█████████▋| 35/36 [00:09<00:00, 3.81it/s]\n100%|██████████| 36/36 [00:09<00:00, 3.81it/s]\n100%|██████████| 36/36 [00:09<00:00, 3.82it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 10.287849508, "total_time": 10.30389 }, "output": [ "https://replicate.delivery/xezq/GXxPh6YWYeT7QyaZdsl5eyZew9ZhsbbT9HwFtQsizQM0jzIpA/out-0.webp" ], "started_at": "2025-04-19T18:24:47.848040Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-azc6vfmbf6c2rqda53p47bqjv6wjz7ybdx5mpezajbvtpljnmjsq", "get": "https://api.replicate.com/v1/predictions/3qpexhp4v1rm80cp9x5srjfrvc", "cancel": "https://api.replicate.com/v1/predictions/3qpexhp4v1rm80cp9x5srjfrvc/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.58s Using seed: 9412 Prompt: Subject: Full-body portrait of TOKMAHI standing poised on a dimly lit New York street corner, 1980s–1990s style, wearing an elegant, structured women's jacket with sharp tailoring—shoulders cinched, collar exaggerated, silhouette sculpted like modern architecture; her afro styled into a symmetrical halo with fine sculptural lines Style: avant-garde deconstructive elegance Composition: full-body in frame, blended with fragmented glass shards and urban textures—subway tiles, neon signs, steel fire escapes, graffiti‑tagged newspaper boxes; background tilted subtly for surreal depth Lighting: high-contrast rim lighting outlining her form, layered with ambient glow from flickering neon signage and backlit storefront windows Color Palette: dramatic blacks and muted grays punctuated by electric cyan, fuchsia, brass gold, and washed denim tones Mood/Atmosphere: commanding, mysterious, regal within the chaos—avant-garde elegance clashing with retro NYC grit Technical Details: 50mm tilt-shift lens simulation, f/2.0 shallow depth of field, analog film grain (Kodak Elite Chrome vibe), gentle motion blur on passing taxis and distant crowds Additional Elements: glitch trails from nearby traffic lights, wind-tousled fabric motion, reflective puddles mirroring her form, subtle holographic flares across glass debris --ar 3:4 --v flux1 --q 2 --style raw --chaos 80 [!] txt2img mode 0%| | 0/36 [00:00<?, ?it/s] 3%|▎ | 1/36 [00:00<00:09, 3.69it/s] 6%|▌ | 2/36 [00:00<00:08, 4.08it/s] 8%|▊ | 3/36 [00:00<00:08, 3.95it/s] 11%|█ | 4/36 [00:01<00:08, 3.90it/s] 14%|█▍ | 5/36 [00:01<00:08, 3.87it/s] 17%|█▋ | 6/36 [00:01<00:07, 3.85it/s] 19%|█▉ | 7/36 [00:01<00:07, 3.84it/s] 22%|██▏ | 8/36 [00:02<00:07, 3.83it/s] 25%|██▌ | 9/36 [00:02<00:07, 3.83it/s] 28%|██▊ | 10/36 [00:02<00:06, 3.83it/s] 31%|███ | 11/36 [00:02<00:06, 3.82it/s] 33%|███▎ | 12/36 [00:03<00:06, 3.82it/s] 36%|███▌ | 13/36 [00:03<00:06, 3.83it/s] 39%|███▉ | 14/36 [00:03<00:05, 3.82it/s] 42%|████▏ | 15/36 [00:03<00:05, 3.81it/s] 44%|████▍ | 16/36 [00:04<00:05, 3.82it/s] 47%|████▋ | 17/36 [00:04<00:04, 3.82it/s] 50%|█████ | 18/36 [00:04<00:04, 3.80it/s] 53%|█████▎ | 19/36 [00:04<00:04, 3.81it/s] 56%|█████▌ | 20/36 [00:05<00:04, 3.81it/s] 58%|█████▊ | 21/36 [00:05<00:03, 3.81it/s] 61%|██████ | 22/36 [00:05<00:03, 3.81it/s] 64%|██████▍ | 23/36 [00:06<00:03, 3.81it/s] 67%|██████▋ | 24/36 [00:06<00:03, 3.81it/s] 69%|██████▉ | 25/36 [00:06<00:02, 3.81it/s] 72%|███████▏ | 26/36 [00:06<00:02, 3.80it/s] 75%|███████▌ | 27/36 [00:07<00:02, 3.80it/s] 78%|███████▊ | 28/36 [00:07<00:02, 3.81it/s] 81%|████████ | 29/36 [00:07<00:01, 3.81it/s] 83%|████████▎ | 30/36 [00:07<00:01, 3.80it/s] 86%|████████▌ | 31/36 [00:08<00:01, 3.80it/s] 89%|████████▉ | 32/36 [00:08<00:01, 3.81it/s] 92%|█████████▏| 33/36 [00:08<00:00, 3.81it/s] 94%|█████████▍| 34/36 [00:08<00:00, 3.80it/s] 97%|█████████▋| 35/36 [00:09<00:00, 3.81it/s] 100%|██████████| 36/36 [00:09<00:00, 3.81it/s] 100%|██████████| 36/36 [00:09<00:00, 3.82it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDzmv10jkw69rmc0cp9xsajge3s8StatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s face and shoulders in stark monochrome Style: film noir Composition: tight crop, slight head turn away from light Lighting: strong side light with deep shadows, “Venetian blind” pattern optional Color Palette: pure black & white, no grays Mood/Atmosphere: brooding, suspenseful intrigue Technical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain Additional Elements: subtle smoke curl, reflective glint in her eye --ar 1:1 --v flux1 --q 2 --style raw --chaos 25
- go_fast
- lora_scale
- 1.03
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.12
- output_quality
- 80
- prompt_strength
- 1
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { model: "dev", prompt: "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", go_fast: false, lora_scale: 1.03, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.12, output_quality: 80, prompt_strength: 1, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": False, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'model="dev"' \ -i $'prompt="\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.03' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.12' \ -i 'output_quality=80' \ -i 'prompt_strength=1' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "model": "dev", "prompt": "\\nSubject: TOKMAHI’s face and shoulders in stark monochrome \\nStyle: film noir \\nComposition: tight crop, slight head turn away from light \\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \\nColor Palette: pure black & white, no grays \\nMood/Atmosphere: brooding, suspenseful intrigue \\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \\nAdditional Elements: subtle smoke curl, reflective glint in her eye \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \\n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:07:13.902149Z", "created_at": "2025-04-19T19:07:05.138000Z", "data_removed": false, "error": null, "id": "zmv10jkw69rmc0cp9xsajge3s8", "input": { "model": "dev", "prompt": "\nSubject: TOKMAHI’s face and shoulders in stark monochrome \nStyle: film noir \nComposition: tight crop, slight head turn away from light \nLighting: strong side light with deep shadows, “Venetian blind” pattern optional \nColor Palette: pure black & white, no grays \nMood/Atmosphere: brooding, suspenseful intrigue \nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain \nAdditional Elements: subtle smoke curl, reflective glint in her eye \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25 \n", "go_fast": false, "lora_scale": 1.03, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.12, "output_quality": 80, "prompt_strength": 1, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.57s\nUsing seed: 26655\nPrompt:\nSubject: TOKMAHI’s face and shoulders in stark monochrome\nStyle: film noir\nComposition: tight crop, slight head turn away from light\nLighting: strong side light with deep shadows, “Venetian blind” pattern optional\nColor Palette: pure black & white, no grays\nMood/Atmosphere: brooding, suspenseful intrigue\nTechnical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain\nAdditional Elements: subtle smoke curl, reflective glint in her eye\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 25\n[!] txt2img mode\n 0%| | 0/30 [00:00<?, ?it/s]\n 3%|▎ | 1/30 [00:00<00:07, 3.74it/s]\n 7%|▋ | 2/30 [00:00<00:06, 4.14it/s]\n 10%|█ | 3/30 [00:00<00:06, 4.00it/s]\n 13%|█▎ | 4/30 [00:01<00:06, 3.91it/s]\n 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s]\n 20%|██ | 6/30 [00:01<00:06, 3.86it/s]\n 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s]\n 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s]\n 30%|███ | 9/30 [00:02<00:05, 3.81it/s]\n 33%|███▎ | 10/30 [00:02<00:05, 3.82it/s]\n 37%|███▋ | 11/30 [00:02<00:04, 3.82it/s]\n 40%|████ | 12/30 [00:03<00:04, 3.81it/s]\n 43%|████▎ | 13/30 [00:03<00:04, 3.81it/s]\n 47%|████▋ | 14/30 [00:03<00:04, 3.81it/s]\n 50%|█████ | 15/30 [00:03<00:03, 3.81it/s]\n 53%|█████▎ | 16/30 [00:04<00:03, 3.81it/s]\n 57%|█████▋ | 17/30 [00:04<00:03, 3.81it/s]\n 60%|██████ | 18/30 [00:04<00:03, 3.82it/s]\n 63%|██████▎ | 19/30 [00:04<00:02, 3.81it/s]\n 67%|██████▋ | 20/30 [00:05<00:02, 3.81it/s]\n 70%|███████ | 21/30 [00:05<00:02, 3.81it/s]\n 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s]\n 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s]\n 80%|████████ | 24/30 [00:06<00:01, 3.80it/s]\n 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s]\n 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s]\n 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s]\n 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s]\n 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.81it/s]\n100%|██████████| 30/30 [00:07<00:00, 3.83it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 8.683775321, "total_time": 8.764149 }, "output": [ "https://replicate.delivery/xezq/mf4Wl2W6330tCirAziy1J332uaKGfukb15h8NJS34gYhZakUA/out-0.webp" ], "started_at": "2025-04-19T19:07:05.218374Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-7hzldo5mobtamli6wxshakqdiq673m3yuim7a5hialfbrrssydgq", "get": "https://api.replicate.com/v1/predictions/zmv10jkw69rmc0cp9xsajge3s8", "cancel": "https://api.replicate.com/v1/predictions/zmv10jkw69rmc0cp9xsajge3s8/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.57s Using seed: 26655 Prompt: Subject: TOKMAHI’s face and shoulders in stark monochrome Style: film noir Composition: tight crop, slight head turn away from light Lighting: strong side light with deep shadows, “Venetian blind” pattern optional Color Palette: pure black & white, no grays Mood/Atmosphere: brooding, suspenseful intrigue Technical Details: 50 mm lens simulation, f/1.8, high contrast, analog film grain Additional Elements: subtle smoke curl, reflective glint in her eye --ar 1:1 --v flux1 --q 2 --style raw --chaos 25 [!] txt2img mode 0%| | 0/30 [00:00<?, ?it/s] 3%|▎ | 1/30 [00:00<00:07, 3.74it/s] 7%|▋ | 2/30 [00:00<00:06, 4.14it/s] 10%|█ | 3/30 [00:00<00:06, 4.00it/s] 13%|█▎ | 4/30 [00:01<00:06, 3.91it/s] 17%|█▋ | 5/30 [00:01<00:06, 3.87it/s] 20%|██ | 6/30 [00:01<00:06, 3.86it/s] 23%|██▎ | 7/30 [00:01<00:05, 3.85it/s] 27%|██▋ | 8/30 [00:02<00:05, 3.83it/s] 30%|███ | 9/30 [00:02<00:05, 3.81it/s] 33%|███▎ | 10/30 [00:02<00:05, 3.82it/s] 37%|███▋ | 11/30 [00:02<00:04, 3.82it/s] 40%|████ | 12/30 [00:03<00:04, 3.81it/s] 43%|████▎ | 13/30 [00:03<00:04, 3.81it/s] 47%|████▋ | 14/30 [00:03<00:04, 3.81it/s] 50%|█████ | 15/30 [00:03<00:03, 3.81it/s] 53%|█████▎ | 16/30 [00:04<00:03, 3.81it/s] 57%|█████▋ | 17/30 [00:04<00:03, 3.81it/s] 60%|██████ | 18/30 [00:04<00:03, 3.82it/s] 63%|██████▎ | 19/30 [00:04<00:02, 3.81it/s] 67%|██████▋ | 20/30 [00:05<00:02, 3.81it/s] 70%|███████ | 21/30 [00:05<00:02, 3.81it/s] 73%|███████▎ | 22/30 [00:05<00:02, 3.80it/s] 77%|███████▋ | 23/30 [00:06<00:01, 3.81it/s] 80%|████████ | 24/30 [00:06<00:01, 3.80it/s] 83%|████████▎ | 25/30 [00:06<00:01, 3.81it/s] 87%|████████▋ | 26/30 [00:06<00:01, 3.81it/s] 90%|█████████ | 27/30 [00:07<00:00, 3.81it/s] 93%|█████████▎| 28/30 [00:07<00:00, 3.80it/s] 97%|█████████▋| 29/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.81it/s] 100%|██████████| 30/30 [00:07<00:00, 3.83it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDnbcp2e9yc5rmc0cp9y689rz10wStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.22
- output_quality
- 80
- prompt_strength
- 0.5
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", model: "dev", prompt: "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.22, output_quality: 80, prompt_strength: 0.5, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.22' \ -i 'output_quality=80' \ -i 'prompt_strength=0.5' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:35:18.804830Z", "created_at": "2025-04-19T19:35:13.249000Z", "data_removed": false, "error": null, "id": "nbcp2e9yc5rmc0cp9y689rz10w", "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.55s\nUsing seed: 55281\nPrompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue\nStyle: vintage film\nComposition: 3/4 crop from shoulders up, slight head tilt\nLighting: warm golden‑hour glow with gentle side fill\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia\nMood/Atmosphere: nostalgic, timeless warmth\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette\nAdditional Elements: subtle light leaks, soft border fade\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20\nInput image size: 1024x1024\n[!] Resizing input image from 1024x1024 to 1024x1024\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:03, 4.47it/s]\n 13%|█▎ | 2/15 [00:00<00:03, 4.08it/s]\n 20%|██ | 3/15 [00:00<00:03, 3.96it/s]\n 27%|██▋ | 4/15 [00:01<00:02, 3.90it/s]\n 33%|███▎ | 5/15 [00:01<00:02, 3.87it/s]\n 40%|████ | 6/15 [00:01<00:02, 3.85it/s]\n 47%|████▋ | 7/15 [00:01<00:02, 3.85it/s]\n 53%|█████▎ | 8/15 [00:02<00:01, 3.84it/s]\n 60%|██████ | 9/15 [00:02<00:01, 3.83it/s]\n 67%|██████▋ | 10/15 [00:02<00:01, 3.83it/s]\n 73%|███████▎ | 11/15 [00:02<00:01, 3.82it/s]\n 80%|████████ | 12/15 [00:03<00:00, 3.82it/s]\n 87%|████████▋ | 13/15 [00:03<00:00, 3.82it/s]\n 93%|█████████▎| 14/15 [00:03<00:00, 3.83it/s]\n100%|██████████| 15/15 [00:03<00:00, 3.82it/s]\n100%|██████████| 15/15 [00:03<00:00, 3.86it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 4.943164975, "total_time": 5.55583 }, "output": [ "https://replicate.delivery/xezq/w6I05cTlBfRkY62zgxN9JlCHyMsqz7amMH1xXjfe6Khsn1IpA/out-0.webp" ], "started_at": "2025-04-19T19:35:13.861665Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-zzhejkjkeiu7xf6hn22ydi62tlj64yowpcf6dmox7tgsqeulmnwq", "get": "https://api.replicate.com/v1/predictions/nbcp2e9yc5rmc0cp9y689rz10w", "cancel": "https://api.replicate.com/v1/predictions/nbcp2e9yc5rmc0cp9y689rz10w/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.55s Using seed: 55281 Prompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20 Input image size: 1024x1024 [!] Resizing input image from 1024x1024 to 1024x1024 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:03, 4.47it/s] 13%|█▎ | 2/15 [00:00<00:03, 4.08it/s] 20%|██ | 3/15 [00:00<00:03, 3.96it/s] 27%|██▋ | 4/15 [00:01<00:02, 3.90it/s] 33%|███▎ | 5/15 [00:01<00:02, 3.87it/s] 40%|████ | 6/15 [00:01<00:02, 3.85it/s] 47%|████▋ | 7/15 [00:01<00:02, 3.85it/s] 53%|█████▎ | 8/15 [00:02<00:01, 3.84it/s] 60%|██████ | 9/15 [00:02<00:01, 3.83it/s] 67%|██████▋ | 10/15 [00:02<00:01, 3.83it/s] 73%|███████▎ | 11/15 [00:02<00:01, 3.82it/s] 80%|████████ | 12/15 [00:03<00:00, 3.82it/s] 87%|████████▋ | 13/15 [00:03<00:00, 3.82it/s] 93%|█████████▎| 14/15 [00:03<00:00, 3.83it/s] 100%|██████████| 15/15 [00:03<00:00, 3.82it/s] 100%|██████████| 15/15 [00:03<00:00, 3.86it/s] Total safe images: 1 out of 1
Prediction
koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104IDnbcp2e9yc5rmc0cp9y689rz10wStatusSucceededSourceWebHardwareH100Total durationCreatedInput
- model
- dev
- prompt
- Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20
- go_fast
- lora_scale
- 1.1
- megapixels
- 1
- num_outputs
- 1
- aspect_ratio
- 1:1
- output_format
- webp
- guidance_scale
- 2.22
- output_quality
- 80
- prompt_strength
- 0.5
- extra_lora_scale
- 1
- num_inference_steps
- 30
{ "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }
Install Replicate’s Node.js client library:npm install replicate
Import and set up the client:import Replicate from "replicate"; import fs from "node:fs"; const replicate = new Replicate({ auth: process.env.REPLICATE_API_TOKEN, });
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
const output = await replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", { input: { image: "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", model: "dev", prompt: "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", go_fast: false, lora_scale: 1.1, megapixels: "1", num_outputs: 1, aspect_ratio: "1:1", output_format: "webp", guidance_scale: 2.22, output_quality: 80, prompt_strength: 0.5, extra_lora_scale: 1, num_inference_steps: 30 } } ); // To access the file URL: console.log(output[0].url()); //=> "http://example.com" // To write the file to disk: fs.writeFile("my-image.png", output[0]);
To learn more, take a look at the guide on getting started with Node.js.
Install Replicate’s Python client library:pip install replicate
Import the client:import replicate
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
output = replicate.run( "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", input={ "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": False, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } ) # To access the file URL: print(output[0].url()) #=> "http://example.com" # To write the file to disk: with open("my-image.png", "wb") as file: file.write(output[0].read())
To learn more, take a look at the guide on getting started with Python.
Run koolamusic/flux-dev-lora-mahi using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.
curl -s -X POST \ -H "Authorization: Bearer $REPLICATE_API_TOKEN" \ -H "Content-Type: application/json" \ -H "Prefer: wait" \ -d $'{ "version": "koolamusic/flux-dev-lora-mahi:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104", "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ https://api.replicate.com/v1/predictions
To learn more, take a look at Replicate’s HTTP API reference docs.
You can run this model locally using Cog. First, install Cog:brew install cog
If you don’t have Homebrew, there are other installation options available.
Run this to download the model and run it in your local environment:
cog predict r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104 \ -i 'image="https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png"' \ -i 'model="dev"' \ -i $'prompt="Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n"' \ -i 'go_fast=false' \ -i 'lora_scale=1.1' \ -i 'megapixels="1"' \ -i 'num_outputs=1' \ -i 'aspect_ratio="1:1"' \ -i 'output_format="webp"' \ -i 'guidance_scale=2.22' \ -i 'output_quality=80' \ -i 'prompt_strength=0.5' \ -i 'extra_lora_scale=1' \ -i 'num_inference_steps=30'
To learn more, take a look at the Cog documentation.
Run this to download the model and run it in your local environment:
docker run -d -p 5000:5000 --gpus=all r8.im/koolamusic/flux-dev-lora-mahi@sha256:99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104
curl -s -X POST \ -H "Content-Type: application/json" \ -d $'{ "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \\nStyle: vintage film \\nComposition: 3/4 crop from shoulders up, slight head tilt \\nLighting: warm golden‑hour glow with gentle side fill \\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \\nMood/Atmosphere: nostalgic, timeless warmth \\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \\nAdditional Elements: subtle light leaks, soft border fade \\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \\n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 } }' \ http://localhost:5000/predictions
To learn more, take a look at the Cog documentation.
Output
{ "completed_at": "2025-04-19T19:35:18.804830Z", "created_at": "2025-04-19T19:35:13.249000Z", "data_removed": false, "error": null, "id": "nbcp2e9yc5rmc0cp9y689rz10w", "input": { "image": "https://replicate.delivery/pbxt/Mrjb2syYbjTRAagi6muWrZOi1hxeP4Hypr8hBQIrDdNUCUOi/ChatGPT%20Image%20Apr%2019%2C%202025%20at%2007_34_43%20PM.png", "model": "dev", "prompt": "Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue \nStyle: vintage film \nComposition: 3/4 crop from shoulders up, slight head tilt \nLighting: warm golden‑hour glow with gentle side fill \nColor Palette: faded pastels—soft yellows, washed pinks, light sepia \nMood/Atmosphere: nostalgic, timeless warmth \nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette \nAdditional Elements: subtle light leaks, soft border fade \n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20 \n", "go_fast": false, "lora_scale": 1.1, "megapixels": "1", "num_outputs": 1, "aspect_ratio": "1:1", "output_format": "webp", "guidance_scale": 2.22, "output_quality": 80, "prompt_strength": 0.5, "extra_lora_scale": 1, "num_inference_steps": 30 }, "logs": "Loaded LoRAs in 0.55s\nUsing seed: 55281\nPrompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue\nStyle: vintage film\nComposition: 3/4 crop from shoulders up, slight head tilt\nLighting: warm golden‑hour glow with gentle side fill\nColor Palette: faded pastels—soft yellows, washed pinks, light sepia\nMood/Atmosphere: nostalgic, timeless warmth\nTechnical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette\nAdditional Elements: subtle light leaks, soft border fade\n--ar 1:1 --v flux1 --q 2 --style raw --chaos 20\nInput image size: 1024x1024\n[!] Resizing input image from 1024x1024 to 1024x1024\n[!] img2img mode\n 0%| | 0/15 [00:00<?, ?it/s]\n 7%|▋ | 1/15 [00:00<00:03, 4.47it/s]\n 13%|█▎ | 2/15 [00:00<00:03, 4.08it/s]\n 20%|██ | 3/15 [00:00<00:03, 3.96it/s]\n 27%|██▋ | 4/15 [00:01<00:02, 3.90it/s]\n 33%|███▎ | 5/15 [00:01<00:02, 3.87it/s]\n 40%|████ | 6/15 [00:01<00:02, 3.85it/s]\n 47%|████▋ | 7/15 [00:01<00:02, 3.85it/s]\n 53%|█████▎ | 8/15 [00:02<00:01, 3.84it/s]\n 60%|██████ | 9/15 [00:02<00:01, 3.83it/s]\n 67%|██████▋ | 10/15 [00:02<00:01, 3.83it/s]\n 73%|███████▎ | 11/15 [00:02<00:01, 3.82it/s]\n 80%|████████ | 12/15 [00:03<00:00, 3.82it/s]\n 87%|████████▋ | 13/15 [00:03<00:00, 3.82it/s]\n 93%|█████████▎| 14/15 [00:03<00:00, 3.83it/s]\n100%|██████████| 15/15 [00:03<00:00, 3.82it/s]\n100%|██████████| 15/15 [00:03<00:00, 3.86it/s]\nTotal safe images: 1 out of 1", "metrics": { "predict_time": 4.943164975, "total_time": 5.55583 }, "output": [ "https://replicate.delivery/xezq/w6I05cTlBfRkY62zgxN9JlCHyMsqz7amMH1xXjfe6Khsn1IpA/out-0.webp" ], "started_at": "2025-04-19T19:35:13.861665Z", "status": "succeeded", "urls": { "stream": "https://stream.replicate.com/v1/files/bcwr-zzhejkjkeiu7xf6hn22ydi62tlj64yowpcf6dmox7tgsqeulmnwq", "get": "https://api.replicate.com/v1/predictions/nbcp2e9yc5rmc0cp9y689rz10w", "cancel": "https://api.replicate.com/v1/predictions/nbcp2e9yc5rmc0cp9y689rz10w/cancel" }, "version": "99d540469709ade366245974412001257c567ba82f7dc0a7e151bf8326290104" }
Generated inLoaded LoRAs in 0.55s Using seed: 55281 Prompt: Subject: TOKMAHI’s elegant portrait, softly lit in a nostalgic hue Style: vintage film Composition: 3/4 crop from shoulders up, slight head tilt Lighting: warm golden‑hour glow with gentle side fill Color Palette: faded pastels—soft yellows, washed pinks, light sepia Mood/Atmosphere: nostalgic, timeless warmth Technical Details: 50 mm lens simulation, f/2.0, Kodachrome film emulation, light grain and vignette Additional Elements: subtle light leaks, soft border fade --ar 1:1 --v flux1 --q 2 --style raw --chaos 20 Input image size: 1024x1024 [!] Resizing input image from 1024x1024 to 1024x1024 [!] img2img mode 0%| | 0/15 [00:00<?, ?it/s] 7%|▋ | 1/15 [00:00<00:03, 4.47it/s] 13%|█▎ | 2/15 [00:00<00:03, 4.08it/s] 20%|██ | 3/15 [00:00<00:03, 3.96it/s] 27%|██▋ | 4/15 [00:01<00:02, 3.90it/s] 33%|███▎ | 5/15 [00:01<00:02, 3.87it/s] 40%|████ | 6/15 [00:01<00:02, 3.85it/s] 47%|████▋ | 7/15 [00:01<00:02, 3.85it/s] 53%|█████▎ | 8/15 [00:02<00:01, 3.84it/s] 60%|██████ | 9/15 [00:02<00:01, 3.83it/s] 67%|██████▋ | 10/15 [00:02<00:01, 3.83it/s] 73%|███████▎ | 11/15 [00:02<00:01, 3.82it/s] 80%|████████ | 12/15 [00:03<00:00, 3.82it/s] 87%|████████▋ | 13/15 [00:03<00:00, 3.82it/s] 93%|█████████▎| 14/15 [00:03<00:00, 3.83it/s] 100%|██████████| 15/15 [00:03<00:00, 3.82it/s] 100%|██████████| 15/15 [00:03<00:00, 3.86it/s] Total safe images: 1 out of 1
Want to make some of these yourself?
Run this model