Wan I2V 480p LoRA
Wan-2.1 i2v model with LoRA, generate high-quality videos with superior visual quality and motion diversity
Features
Wan2.1-i2v-480p-lora is an open-source AI video generation model developed by Alibaba Cloud, designed for image-to-video tasks. It incorporates Low-Rank Adaptation (LoRA) techniques, enabling efficient fine-tuning of pre-trained models to generate videos with specified effects from reference images. The 14-billion-parameter professional version excels in generating complex motions and simulating physical dynamics, delivering exceptional performance. Built upon a causal 3D Variational Autoencoder (VAE) and Video Diffusion Transformer architecture, Wan2.1-i2v-480p-lora efficiently models spatiotemporal dependencies. In the authoritative VBench evaluation, the 14B version achieved a leading score of 86.22%, surpassing models like Sora, Luma, and Pika, and securing the top position. The model is available on Wavespeed AI, providing convenient access for developers.
Key Features
- Exceptional Image-to-Video Performance: Specially optimized for converting images into videos, WAN-2.1/i2v-480p-lora delivers state-of-the-art visual quality and natural motion dynamics.
- Trained on the Wan2.1 14B 480p I2V base model.
- Consistent results across different object types .
- Simple prompt structure that's easy to adapt.
- Squish :Transform any image into a video of it being rotated.
- Rotate : Transform any image into a video of it being squished.
- Cakeify:Transform any image into a video of it being cakeified.
- Crush:Transform any image into a video of it being crushed.
- Painting:Transform any image into a video of the subject as a historic painting version of themselves!
- Deflate:Transform any image into a video of it being deflated.
- Optimized for Speed: Leveraging advanced inference acceleration techniques from WaveSpeedAI, the system minimizes computational overhead and latency, ensuring rapid video generation without sacrificing output quality.
ComfyUI
wan-2.1/i2v-480p-lora is also available on ComfyUI, providing local inference capabilities through a node-based workflow, ensuring flexible and efficient image generation on your system.
Limitations
- Creative Focus: WAN-2.1/i2v-480p-lora is designed exclusively for creative image-to-video transformation and is not intended for generating factually accurate content.
- Statistical Biases: As with other data-driven models, there is potential for inadvertently propagating biases found in its training data.
- Prompt Sensitivity: The model’s performance is influenced by the clarity and precision of the input image; less detailed images may lead to variations in output quality.
- Scope of Functionality: This model is solely focused on image-to-video generation and does not support other video generative tasks such as text-to-video or video editing.
Out-of-Scope Use
The model and its derivatives may not be used in any way that violates applicable national, federal, state, local, or international law or regulation, including but not limited to:
- Exploiting, harming, or attempting to exploit or harm minors, including solicitation, creation, acquisition, or dissemination of child exploitative content.
- Generating or disseminating verifiably false information with the intent to harm others.
- Creating or distributing personal identifiable information that could be used to harm an individual.
- Harassing, abusing, threatening, stalking, or bullying individuals or groups.
- Producing non-consensual nudity or illegal pornographic content.
- Making fully automated decisions that adversely affect an individual’s legal rights or create binding obligations.
- Facilitating large-scale disinformation campaigns.
Accelerated Inference
Our accelerated inference approach leverages advanced optimization technology from WavespeedAI. This innovative fusion technique significantly reduces computational overhead and latency, enabling rapid image generation without compromising quality. The entire system is designed to efficiently handle large-scale inference tasks while ensuring that real-time applications achieve an optimal balance between speed and accuracy. For further details, please refer to the blog post.
Authentication
For authentication details, please refer to the Authentication Guide.
API Endpoints
Submit Task & Query Result
# Submit the task
curl --location --request POST "https://api.wavespeed.ai/api/v3/wavespeed-ai/wan-2.1/i2v-480p-lora" \
--header "Content-Type: application/json" \
--header "Authorization: Bearer ${WAVESPEED_API_KEY}" \
--data-raw '{
"image": "https://d2g64w682n9w0w.cloudfront.net/media/images/1745077093052414125_462ZVSNL.jpg",
"prompt": "The video opens with a cat woman. As the video progresses, the cat woman begins to d3d1at3 deflate it, gradually shrinking and losing shape, eventually flattening completely into a lifeless, deflated mass on the ground",
"negative_prompt": "",
"loras": [
{
"path": "Remade-AI/Deflate",
"scale": 1
}
],
"size": "832*480",
"num_inference_steps": 30,
"duration": 5,
"guidance_scale": 5,
"flow_shift": 3,
"seed": -1,
"enable_safety_checker": true
}'
# Get the result
curl --location --request GET "https://api.wavespeed.ai/api/v3/predictions/${requestId}/result" \
--header "Authorization: Bearer ${WAVESPEED_API_KEY}"
Parameters
Task Submission Parameters
Request Parameters
Parameter | Type | Required | Default | Range | Description |
---|---|---|---|---|---|
image | string | Yes | https://d2g64w682n9w0w.cloudfront.net/media/images/1745077093052414125_462ZVSNL.jpg | - | The image for generating the output. |
prompt | string | Yes | The video opens with a cat woman. As the video progresses, the cat woman begins to d3d1at3 deflate it, gradually shrinking and losing shape, eventually flattening completely into a lifeless, deflated mass on the ground | - | The prompt for generating the output. |
negative_prompt | string | No | - | - | The negative prompt for generating the output. |
loras | array | No | [] | max 3 items | The LoRA weights for generating the output. |
loras[].path | string | Yes | - | Path to the LoRA model | |
loras[].scale | float | Yes | - | 0.0 ~ 4.0 | Scale of the LoRA model |
size | string | No | 832*480 | 832*480, 480*832 | The size of the output. |
num_inference_steps | integer | No | 30 | 1 ~ 40 | The number of inference steps. |
duration | integer | No | 5 | 5 ~ 10 | Generate video duration length seconds. |
guidance_scale | number | No | 5 | 1.01 ~ 10.00 | The guidance scale for generation. |
flow_shift | number | No | 3 | 1.0 ~ 10.0 | The shift value for the timestep schedule for flow matching. |
seed | integer | No | -1 | -1 ~ 9999999999 | The seed for random number generation. |
enable_safety_checker | boolean | No | true | - | Whether to enable the safety checker. |
Response Parameters
Parameter | Type | Description |
---|---|---|
code | integer | HTTP status code (e.g., 200 for success) |
message | string | Status message (e.g., “success”) |
data.id | string | Unique identifier for the prediction, Task Id |
data.model | string | Model ID used for the prediction |
data.outputs | array | Array of URLs to the generated content (empty when status is not completed ) |
data.urls | object | Object containing related API endpoints |
data.urls.get | string | URL to retrieve the prediction result |
data.has_nsfw_contents | array | Array of boolean values indicating NSFW detection for each output |
data.status | string | Status of the task: created , processing , completed , or failed |
data.created_at | string | ISO timestamp of when the request was created (e.g., “2023-04-01T12:34:56.789Z”) |
data.error | string | Error message (empty if no error occurred) |
data.timings | object | Object containing timing details |
data.timings.inference | integer | Inference time in milliseconds |
Result Query Parameters
Result Request Parameters
Parameter | Type | Required | Default | Description |
---|---|---|---|---|
id | string | Yes | - | Task ID |
Result Response Parameters
Parameter | Type | Description |
---|---|---|
code | integer | HTTP status code (e.g., 200 for success) |
message | string | Status message (e.g., “success”) |
data | object | The prediction data object containing all details |
data.id | string | Unique identifier for the prediction |
data.model | string | Model ID used for the prediction |
data.outputs | array | Array of URLs to the generated content (empty when status is not completed ) |
data.urls | object | Object containing related API endpoints |
data.urls.get | string | URL to retrieve the prediction result |
data.has_nsfw_contents | array | Array of boolean values indicating NSFW detection for each output |
data.status | string | Status of the task: created , processing , completed , or failed |
data.created_at | string | ISO timestamp of when the request was created (e.g., “2023-04-01T12:34:56.789Z”) |
data.error | string | Error message (empty if no error occurred) |
data.timings | object | Object containing timing details |
data.timings.inference | integer | Inference time in milliseconds |