aleksa-codes/flux-ghibsky-illustration – Run with an API on Replicate

aleksa-codes / flux-ghibsky-illustration

Flux LoRA, use 'GHIBSKY style' to trigger generation, creates serene and enchanting landscapes with vibrant, surreal skies and intricate, Ghibli-inspired elements reminiscent of the atmospheric beauty found in Makoto Shinkai's works

Warm

Public
74K runs
H100
Weights

Run with an API

Playground API Examples README Versions

Input

prompt

*string

Shift + Return to add a new line

GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lightsGHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights

Prompt for generated image. If you include the `trigger_word` used in the training process you are more likely to activate the trained object, style, or concept in the resulting image.

image

file

Input image for image to image or inpainting mode. If provided, aspect_ratio, width, and height inputs are ignored.

mask

file

Image mask for image inpainting mode. If provided, aspect_ratio, width, and height inputs are ignored.

aspect_ratio

string

Aspect ratio for the generated image. If custom is selected, uses height and width below & will run in bf16 mode

Default: "1:1"

height

integer

(minimum: 256, maximum: 1440)

Height of generated image. Only works if `aspect_ratio` is set to custom. Will be rounded to nearest multiple of 16. Incompatible with fast generation

width

integer

(minimum: 256, maximum: 1440)

Width of generated image. Only works if `aspect_ratio` is set to custom. Will be rounded to nearest multiple of 16. Incompatible with fast generation

prompt_strength

number

(minimum: 0, maximum: 1)

Prompt strength when using img2img. 1.0 corresponds to full destruction of information in image

Default: 0.8

model

string

Which model to run inference with. The dev model performs best with around 28 inference steps but the schnell model only needs 4 steps.

Default: "dev"

num_outputs

integer

(minimum: 1, maximum: 4)

Number of outputs to generate

Default: 1

num_inference_steps

integer

(minimum: 1, maximum: 50)

Number of denoising steps. More steps can give more detailed images, but take longer.

Default: 28

guidance_scale

number

(minimum: 0, maximum: 10)

Guidance scale for the diffusion process. Lower values can give more realistic images. Good values to try are 2, 2.5, 3 and 3.5

Default: 3

seed

integer

Random seed. Set for reproducible generation

output_format

string

Format of the output images

Default: "webp"

output_quality

integer

(minimum: 0, maximum: 100)

Quality when saving the output images, from 0 to 100. 100 is best quality, 0 is lowest quality. Not relevant for .png outputs

Default: 80

disable_safety_checker

boolean

Disable safety checker for generated images.

Default: false

go_fast

boolean

Run faster predictions with model optimized for speed (currently fp8 quantized); disable to run in original bf16

Default: false

megapixels

string

Approximate number of megapixels for generated image

Default: "1"

lora_scale

number

(minimum: -1, maximum: 3)

Determines how strongly the main LoRA should be applied. Sane results between 0 and 1 for base inference. For go_fast we apply a 1.5x multiplier to this value; we've generally seen good performance when scaling the base value by that amount. You may still need to experiment to find the best value for your particular lora.

Default: 1

extra_lora

string

Shift + Return to add a new line

Load LoRA weights. Supports Replicate models in the format <owner>/<username> or <owner>/<username>/<version>, HuggingFace URLs in the format huggingface.co/<owner>/<model-name>, CivitAI URLs in the format civitai.com/models/<id>[/<model-name>], or arbitrary .safetensors URLs from the Internet. For example, 'fofr/flux-pixar-cars'

extra_lora_scale

number

(minimum: -1, maximum: 3)

Determines how strongly the extra LoRA should be applied. Sane results between 0 and 1 for base inference. For go_fast we apply a 1.5x multiplier to this value; we've generally seen good performance when scaling the base value by that amount. You may still need to experiment to find the best value for your particular lora.

Default: 1

Run this model in Node.js with one line of code:

npx create-replicate --model=aleksa-codes/flux-ghibsky-illustration

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run aleksa-codes/flux-ghibsky-illustration using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "aleksa-codes/flux-ghibsky-illustration:a9f94946fa0377091ac0bcfe61b0d62ad9a85224e4b421b677d4747914b908c0",
  {
    input: {
      model: "dev",
      prompt: "GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights",
      go_fast: false,
      lora_scale: 1,
      megapixels: "1",
      num_outputs: 1,
      aspect_ratio: "9:16",
      output_format: "jpg",
      guidance_scale: 3.5,
      output_quality: 100,
      prompt_strength: 0.8,
      extra_lora_scale: 1,
      num_inference_steps: 28
    }
  }
);
console.log(output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run aleksa-codes/flux-ghibsky-illustration using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "aleksa-codes/flux-ghibsky-illustration:a9f94946fa0377091ac0bcfe61b0d62ad9a85224e4b421b677d4747914b908c0",
    input={
        "model": "dev",
        "prompt": "GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights",
        "go_fast": False,
        "lora_scale": 1,
        "megapixels": "1",
        "num_outputs": 1,
        "aspect_ratio": "9:16",
        "output_format": "jpg",
        "guidance_scale": 3.5,
        "output_quality": 100,
        "prompt_strength": 0.8,
        "extra_lora_scale": 1,
        "num_inference_steps": 28
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run aleksa-codes/flux-ghibsky-illustration 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": "a9f94946fa0377091ac0bcfe61b0d62ad9a85224e4b421b677d4747914b908c0",
    "input": {
      "model": "dev",
      "prompt": "GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights",
      "go_fast": false,
      "lora_scale": 1,
      "megapixels": "1",
      "num_outputs": 1,
      "aspect_ratio": "9:16",
      "output_format": "jpg",
      "guidance_scale": 3.5,
      "output_quality": 100,
      "prompt_strength": 0.8,
      "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.

Output

{
  "completed_at": "2024-08-20T16:50:00.663575Z",
  "created_at": "2024-08-20T16:49:42.138000Z",
  "data_removed": false,
  "error": null,
  "id": "p6amm8xwz9rm00che2rtkya19r",
  "input": {
    "model": "dev",
    "prompt": "GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights",
    "lora_scale": 1,
    "num_outputs": 1,
    "aspect_ratio": "9:16",
    "output_format": "jpg",
    "guidance_scale": 3.5,
    "output_quality": 100,
    "num_inference_steps": 28
  },
  "logs": "Using seed: 64884\nPrompt: GHIBSKY style, a cat on a windowsill gazing out at a starry night sky and distant city lights\ntxt2img mode\nUsing dev model\nLoading LoRA weights\nLoRA weights loaded successfully\n  0%|          | 0/28 [00:00<?, ?it/s]\n  4%|▎         | 1/28 [00:00<00:07,  3.70it/s]\n  7%|▋         | 2/28 [00:00<00:06,  4.24it/s]\n 11%|█         | 3/28 [00:00<00:06,  3.97it/s]\n 14%|█▍        | 4/28 [00:01<00:06,  3.87it/s]\n 18%|█▊        | 5/28 [00:01<00:06,  3.81it/s]\n 21%|██▏       | 6/28 [00:01<00:05,  3.76it/s]\n 25%|██▌       | 7/28 [00:01<00:05,  3.73it/s]\n 29%|██▊       | 8/28 [00:02<00:05,  3.73it/s]\n 32%|███▏      | 9/28 [00:02<00:05,  3.72it/s]\n 36%|███▌      | 10/28 [00:02<00:04,  3.71it/s]\n 39%|███▉      | 11/28 [00:02<00:04,  3.70it/s]\n 43%|████▎     | 12/28 [00:03<00:04,  3.70it/s]\n 46%|████▋     | 13/28 [00:03<00:04,  3.70it/s]\n 50%|█████     | 14/28 [00:03<00:03,  3.70it/s]\n 54%|█████▎    | 15/28 [00:04<00:03,  3.69it/s]\n 57%|█████▋    | 16/28 [00:04<00:03,  3.70it/s]\n 61%|██████    | 17/28 [00:04<00:02,  3.70it/s]\n 64%|██████▍   | 18/28 [00:04<00:02,  3.70it/s]\n 68%|██████▊   | 19/28 [00:05<00:02,  3.69it/s]\n 71%|███████▏  | 20/28 [00:05<00:02,  3.70it/s]\n 75%|███████▌  | 21/28 [00:05<00:01,  3.70it/s]\n 79%|███████▊  | 22/28 [00:05<00:01,  3.69it/s]\n 82%|████████▏ | 23/28 [00:06<00:01,  3.69it/s]\n 86%|████████▌ | 24/28 [00:06<00:01,  3.69it/s]\n 89%|████████▉ | 25/28 [00:06<00:00,  3.69it/s]\n 93%|█████████▎| 26/28 [00:06<00:00,  3.69it/s]\n 96%|█████████▋| 27/28 [00:07<00:00,  3.69it/s]\n100%|██████████| 28/28 [00:07<00:00,  3.69it/s]\n100%|██████████| 28/28 [00:07<00:00,  3.72it/s]",
  "metrics": {
    "predict_time": 17.565027229000002,
    "total_time": 18.525575
  },
  "output": [
    "https://replicate.delivery/yhqm/9Z9n5qPvKDonG5057ESfBGDisxdVAG0dfPKRRHAieD3wZPpmA/out-0.jpg"
  ],
  "started_at": "2024-08-20T16:49:43.098548Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/p6amm8xwz9rm00che2rtkya19r",
    "cancel": "https://api.replicate.com/v1/predictions/p6amm8xwz9rm00che2rtkya19r/cancel"
  },
  "version": "a9f94946fa0377091ac0bcfe61b0d62ad9a85224e4b421b677d4747914b908c0"
}

Generated in

17.6 seconds

Tweak it ShareReport

Examples

View more examples

Run time and cost

This model costs approximately $0.030 to run on Replicate, or 33 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia H100 GPU hardware. Predictions typically complete within 20 seconds.

Readme

Description

The Flux Ghibsky Illustration model generates landscapes that blend serene, surreal skies with intricate, Ghibli-inspired details. This fusion of styles creates enchanting scenes that capture the essence of both Ghibli’s whimsical charm and Makoto Shinkai’s atmospheric beauty. Perfect for creating dreamy visuals. Feedback is welcome!

Trigger Words

Use GHIBSKY style to trigger the model’s unique aesthetic. Start your prompt with the trigger word, followed by a description of your scene’s elements, such as nature, skies, houses, roads, villages, etc.

If you are getting too realistic images try adding painting to your prompt, for example: GHIBSKY style painting

Training Details (a9f94946)

Trained Using: ostris/flux-dev-lora-trainer
Number of Images: 35
Trigger Word: GHIBSKY
Auto-captioning: Enabled
Auto-captioning Prefix: “”
Auto-captioning Suffix: “, GHIBSKY style”
Training Steps: 1000
Learning Rate: 0.0004
Batch Size: 1
LoRA Rank: 16

Hugging Face

aleksa-codes/flux-ghibsky-illustration

If you’re training your own LoRA model and need a replacement for LLaVA auto captioning that some LoRA training apps use, try GPT Image Captioner, an open-source tool I created that generates AI-powered descriptions for images. This tool streamlines the auto-captioning process by providing a downloadable zip file with caption .txt files that match your image filenames. It integrates seamlessly with platforms like fal LoRA Trainer and Replicate LoRA Trainer.

The tool now supports Ollama for local inference in addition to OpenAI models, which require your own API key. You can use it as a web app or clone/fork the repository to run it locally. For Ollama integration with the web version, you may need to set up a tunnel like ngrok or allow additional web origins. More information can be found in the project’s README.

Input

Output

Examples

Run time and cost

Readme

Description

Trigger Words

Training Details (a9f94946)

Hugging Face

Related Tools