lucataco/wan2.1-i2v-lora – Run with an API on Replicate

lucataco / wan2.1-i2v-lora

Wan2.1 14B 480p LoRA inference via Diffusers (Work in progress)

Cold

Public
423 runs
A100 (80GB)
GitHub
Weights
License

Iterate in the playground

Run with an API

Playground API Examples README Versions

Input

image

*file

Input image

prompt

*string

Shift + Return to add a new line

In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.

Text prompt describing the desired video effect

negative_prompt

string

Shift + Return to add a new line

low quality, bad quality, blurry, pixelated, watermarklow quality, bad quality, blurry, pixelated, watermark

Text prompt describing what to avoid in the video

Default: "low quality, bad quality, blurry, pixelated, watermark"

lora_url

*string

Shift + Return to add a new line

https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensorshttps://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors

URL to the LoRA weights (HuggingFace or CivitAI)

lora_strength

number

(minimum: 0, maximum: 2)

LoRA effect strength

Default: 1

duration

number

(minimum: 1, maximum: 5)

Video duration in seconds

Default: 3

fps

integer

(minimum: 7, maximum: 30)

Frames per second

Default: 16

guidance_scale

number

(minimum: 1, maximum: 20)

Guidance scale for generation

Default: 5

num_inference_steps

integer

(minimum: 1, maximum: 100)

Number of inference steps

Default: 28

resize_mode

string

Image resizing strategy

Default: "auto"

seed

integer

Random seed for generation

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

npx create-replicate --model=lucataco/wan2.1-i2v-lora

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 lucataco/wan2.1-i2v-lora using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "lucataco/wan2.1-i2v-lora:f130c663a974802a8e5826a49b82064a27d5ffeec8f008ae3195a70e49527a97",
  {
    input: {
      fps: 16,
      image: "https://replicate.delivery/pbxt/Mf7Um7V1nQjebWjOi3RndihR5sK95269LxZDY8s17mqW5jda/dog-1024.png",
      prompt: "In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.",
      duration: 3,
      lora_url: "https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors",
      resize_mode: "auto",
      lora_strength: 1,
      guidance_scale: 5,
      negative_prompt: "low quality, bad quality, blurry, pixelated, watermark",
      num_inference_steps: 40
    }
  }
);

// To access the file URL:
console.log(output.url()); //=> "http://example.com"

// To write the file to disk:
fs.writeFile("my-image.png", 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 lucataco/wan2.1-i2v-lora using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "lucataco/wan2.1-i2v-lora:f130c663a974802a8e5826a49b82064a27d5ffeec8f008ae3195a70e49527a97",
    input={
        "fps": 16,
        "image": "https://replicate.delivery/pbxt/Mf7Um7V1nQjebWjOi3RndihR5sK95269LxZDY8s17mqW5jda/dog-1024.png",
        "prompt": "In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.",
        "duration": 3,
        "lora_url": "https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors",
        "resize_mode": "auto",
        "lora_strength": 1,
        "guidance_scale": 5,
        "negative_prompt": "low quality, bad quality, blurry, pixelated, watermark",
        "num_inference_steps": 40
    }
)
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 lucataco/wan2.1-i2v-lora 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": "lucataco/wan2.1-i2v-lora:f130c663a974802a8e5826a49b82064a27d5ffeec8f008ae3195a70e49527a97",
    "input": {
      "fps": 16,
      "image": "https://replicate.delivery/pbxt/Mf7Um7V1nQjebWjOi3RndihR5sK95269LxZDY8s17mqW5jda/dog-1024.png",
      "prompt": "In the video, a miniature dog is presented. The dog is held in a person\'s hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.",
      "duration": 3,
      "lora_url": "https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors",
      "resize_mode": "auto",
      "lora_strength": 1,
      "guidance_scale": 5,
      "negative_prompt": "low quality, bad quality, blurry, pixelated, watermark",
      "num_inference_steps": 40
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

Output

{
  "completed_at": "2025-03-15T07:09:14.636690Z",
  "created_at": "2025-03-15T07:05:10.835000Z",
  "data_removed": false,
  "error": null,
  "id": "p0xwwxqvpdrj00cnk2fawv6kbr",
  "input": {
    "fps": 16,
    "image": "https://replicate.delivery/pbxt/Mf7Um7V1nQjebWjOi3RndihR5sK95269LxZDY8s17mqW5jda/dog-1024.png",
    "prompt": "In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.",
    "duration": 3,
    "lora_url": "https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors",
    "resize_mode": "auto",
    "lora_strength": 1,
    "guidance_scale": 5,
    "negative_prompt": "low quality, bad quality, blurry, pixelated, watermark",
    "num_inference_steps": 40
  },
  "logs": "Starting prediction with: prompt='In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.', lora_strength=1.0, duration=3.0s\nCalculated 49 frames for 3.0s at 16 fps\nLoading input image from: /tmp/tmpl6fn4nyqdog-1024.png\nImage loaded successfully: 1001x991\nAuto-selected fixed square dimensions: 512x512 for aspect ratio 0.99\nFinal dimensions: 512x512\nDownloading LoRA from: https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors\nDownloading from HuggingFace: repo_id=Remade-AI/Squish, filename=squish_18.safetensors\nLoading LoRA weights from: /root/.cache/huggingface/hub/models--Remade-AI--Squish/snapshots/70e32c7d833743c88352e2ac973968417fb5b051/squish_18.safetensors with strength: 1.0\nLoRA weights loaded successfully\nStarting video generation: 49 frames, 40 steps, guidance=5.0\n  0%|          | 0/40 [00:00<?, ?it/s]\n  2%|▎         | 1/40 [00:05<03:48,  5.87s/it]\n  5%|▌         | 2/40 [00:11<03:31,  5.55s/it]\n  8%|▊         | 3/40 [00:17<03:31,  5.71s/it]\n 10%|█         | 4/40 [00:22<03:28,  5.78s/it]\n 12%|█▎        | 5/40 [00:28<03:23,  5.82s/it]\n 15%|█▌        | 6/40 [00:34<03:19,  5.85s/it]\n 18%|█▊        | 7/40 [00:40<03:13,  5.87s/it]\n 20%|██        | 8/40 [00:46<03:08,  5.89s/it]\n 22%|██▎       | 9/40 [00:52<03:02,  5.90s/it]\n 25%|██▌       | 10/40 [00:58<02:57,  5.91s/it]\n 28%|██▊       | 11/40 [01:04<02:51,  5.91s/it]\n 30%|███       | 12/40 [01:10<02:45,  5.92s/it]\n 32%|███▎      | 13/40 [01:16<02:39,  5.92s/it]\n 35%|███▌      | 14/40 [01:22<02:33,  5.92s/it]\n 38%|███▊      | 15/40 [01:28<02:28,  5.92s/it]\n 40%|████      | 16/40 [01:34<02:22,  5.92s/it]\n 42%|████▎     | 17/40 [01:39<02:16,  5.92s/it]\n 45%|████▌     | 18/40 [01:45<02:10,  5.92s/it]\n 48%|████▊     | 19/40 [01:51<02:04,  5.92s/it]\n 50%|█████     | 20/40 [01:57<01:58,  5.92s/it]\n 52%|█████▎    | 21/40 [02:03<01:52,  5.92s/it]\n 55%|█████▌    | 22/40 [02:09<01:46,  5.92s/it]\n 57%|█████▊    | 23/40 [02:15<01:40,  5.92s/it]\n 60%|██████    | 24/40 [02:21<01:34,  5.92s/it]\n 62%|██████▎   | 25/40 [02:27<01:28,  5.92s/it]\n 65%|██████▌   | 26/40 [02:33<01:22,  5.92s/it]\n 68%|██████▊   | 27/40 [02:39<01:17,  5.92s/it]\n 70%|███████   | 28/40 [02:45<01:11,  5.93s/it]\n 72%|███████▎  | 29/40 [02:51<01:05,  5.92s/it]\n 75%|███████▌  | 30/40 [02:56<00:59,  5.93s/it]\n 78%|███████▊  | 31/40 [03:02<00:53,  5.93s/it]\n 80%|████████  | 32/40 [03:08<00:47,  5.92s/it]\n 82%|████████▎ | 33/40 [03:14<00:41,  5.93s/it]\n 85%|████████▌ | 34/40 [03:20<00:35,  5.92s/it]\n 88%|████████▊ | 35/40 [03:26<00:29,  5.92s/it]\n 90%|█████████ | 36/40 [03:32<00:23,  5.92s/it]\n 92%|█████████▎| 37/40 [03:38<00:17,  5.92s/it]\n 95%|█████████▌| 38/40 [03:44<00:11,  5.92s/it]\n 98%|█████████▊| 39/40 [03:50<00:05,  5.92s/it]\n100%|██████████| 40/40 [03:56<00:00,  5.92s/it]\n100%|██████████| 40/40 [03:56<00:00,  5.90s/it]\nVideo generation completed in 240.27 seconds\nExporting video to: /tmp/tmpbh4k1cz0/output.mp4 at 16 fps\nhuggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...\nTo disable this warning, you can either:\n- Avoid using `tokenizers` before the fork if possible\n- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)\nVideo saved successfully to: /tmp/tmpbh4k1cz0/output.mp4\nCleaned up temporary LoRA file: /root/.cache/huggingface/hub/models--Remade-AI--Squish/snapshots/70e32c7d833743c88352e2ac973968417fb5b051/squish_18.safetensors",
  "metrics": {
    "predict_time": 243.794411078,
    "total_time": 243.80169
  },
  "output": "https://replicate.delivery/yhqm/X0yqkQLeLfuNUESVZHL53DIzmRTtkNJiXFOWsZCPDyKamtYUA/output.mp4",
  "started_at": "2025-03-15T07:05:10.842279Z",
  "status": "succeeded",
  "urls": {
    "stream": "https://stream.replicate.com/v1/files/yswh-stbbisuimfyka4g75i6zoqbo5sk47km22tmjds2v32vdkwqfiowa",
    "get": "https://api.replicate.com/v1/predictions/p0xwwxqvpdrj00cnk2fawv6kbr",
    "cancel": "https://api.replicate.com/v1/predictions/p0xwwxqvpdrj00cnk2fawv6kbr/cancel"
  },
  "version": "f130c663a974802a8e5826a49b82064a27d5ffeec8f008ae3195a70e49527a97"
}

Generated in

4 minutes

Tweak it ShareReport

Starting prediction with: prompt='In the video, a miniature dog is presented. The dog is held in a person's hands. The person then presses on the dog, causing a sq41sh squish effect. The person keeps pressing down on the dog, further showing the sq41sh squish effect.', lora_strength=1.0, duration=3.0s
Calculated 49 frames for 3.0s at 16 fps
Loading input image from: /tmp/tmpl6fn4nyqdog-1024.png
Image loaded successfully: 1001x991
Auto-selected fixed square dimensions: 512x512 for aspect ratio 0.99
Final dimensions: 512x512
Downloading LoRA from: https://huggingface.co/Remade-AI/Squish/resolve/main/squish_18.safetensors
Downloading from HuggingFace: repo_id=Remade-AI/Squish, filename=squish_18.safetensors
Loading LoRA weights from: /root/.cache/huggingface/hub/models--Remade-AI--Squish/snapshots/70e32c7d833743c88352e2ac973968417fb5b051/squish_18.safetensors with strength: 1.0
LoRA weights loaded successfully
Starting video generation: 49 frames, 40 steps, guidance=5.0
  0%|          | 0/40 [00:00<?, ?it/s]
  2%|▎         | 1/40 [00:05<03:48,  5.87s/it]
  5%|▌         | 2/40 [00:11<03:31,  5.55s/it]
  8%|▊         | 3/40 [00:17<03:31,  5.71s/it]
 10%|█         | 4/40 [00:22<03:28,  5.78s/it]
 12%|█▎        | 5/40 [00:28<03:23,  5.82s/it]
 15%|█▌        | 6/40 [00:34<03:19,  5.85s/it]
 18%|█▊        | 7/40 [00:40<03:13,  5.87s/it]
 20%|██        | 8/40 [00:46<03:08,  5.89s/it]
 22%|██▎       | 9/40 [00:52<03:02,  5.90s/it]
 25%|██▌       | 10/40 [00:58<02:57,  5.91s/it]
 28%|██▊       | 11/40 [01:04<02:51,  5.91s/it]
 30%|███       | 12/40 [01:10<02:45,  5.92s/it]
 32%|███▎      | 13/40 [01:16<02:39,  5.92s/it]
 35%|███▌      | 14/40 [01:22<02:33,  5.92s/it]
 38%|███▊      | 15/40 [01:28<02:28,  5.92s/it]
 40%|████      | 16/40 [01:34<02:22,  5.92s/it]
 42%|████▎     | 17/40 [01:39<02:16,  5.92s/it]
 45%|████▌     | 18/40 [01:45<02:10,  5.92s/it]
 48%|████▊     | 19/40 [01:51<02:04,  5.92s/it]
 50%|█████     | 20/40 [01:57<01:58,  5.92s/it]
 52%|█████▎    | 21/40 [02:03<01:52,  5.92s/it]
 55%|█████▌    | 22/40 [02:09<01:46,  5.92s/it]
 57%|█████▊    | 23/40 [02:15<01:40,  5.92s/it]
 60%|██████    | 24/40 [02:21<01:34,  5.92s/it]
 62%|██████▎   | 25/40 [02:27<01:28,  5.92s/it]
 65%|██████▌   | 26/40 [02:33<01:22,  5.92s/it]
 68%|██████▊   | 27/40 [02:39<01:17,  5.92s/it]
 70%|███████   | 28/40 [02:45<01:11,  5.93s/it]
 72%|███████▎  | 29/40 [02:51<01:05,  5.92s/it]
 75%|███████▌  | 30/40 [02:56<00:59,  5.93s/it]
 78%|███████▊  | 31/40 [03:02<00:53,  5.93s/it]
 80%|████████  | 32/40 [03:08<00:47,  5.92s/it]
 82%|████████▎ | 33/40 [03:14<00:41,  5.93s/it]
 85%|████████▌ | 34/40 [03:20<00:35,  5.92s/it]
 88%|████████▊ | 35/40 [03:26<00:29,  5.92s/it]
 90%|█████████ | 36/40 [03:32<00:23,  5.92s/it]
 92%|█████████▎| 37/40 [03:38<00:17,  5.92s/it]
 95%|█████████▌| 38/40 [03:44<00:11,  5.92s/it]
 98%|█████████▊| 39/40 [03:50<00:05,  5.92s/it]
100%|██████████| 40/40 [03:56<00:00,  5.92s/it]
100%|██████████| 40/40 [03:56<00:00,  5.90s/it]
Video generation completed in 240.27 seconds
Exporting video to: /tmp/tmpbh4k1cz0/output.mp4 at 16 fps
huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
- Avoid using `tokenizers` before the fork if possible
- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)
Video saved successfully to: /tmp/tmpbh4k1cz0/output.mp4
Cleaned up temporary LoRA file: /root/.cache/huggingface/hub/models--Remade-AI--Squish/snapshots/70e32c7d833743c88352e2ac973968417fb5b051/squish_18.safetensors

Examples

View more examples

Run time and cost

This model costs approximately $0.47 to run on Replicate, or 2 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 A100 (80GB) GPU hardware. Predictions typically complete within 6 minutes. The predict time for this model varies significantly based on the inputs.

Readme

Work In Progress

Cog model to run Wan2.1 with image to video inference with LoRAs like

Remade-AIs collection