Wan | v2.2 A14B | Image to Video
Animate a still image into a video guided by a text prompt. Features advanced controls including frame interpolation, dual guidance scales, and multiple resolution options.
Quick Start
curl -X POST https://api.eachlabs.ai/v1/prediction \
-H "Content-Type: application/json" \
-H "X-API-Key: $EACHLABS_API_KEY" \
-d '{
"model": "wan-v2-2-a14b-image-to-video",
"version": "0.0.1",
"input": {
"image_url": "https://example.com/landscape.jpg",
"prompt": "Camera slowly pans across the landscape as clouds drift by",
"resolution": "720p",
"aspect_ratio": "16:9",
"num_frames": 81
}
}'
Parameters
| Parameter |
Type |
Default |
Description |
acceleration |
string |
regular |
Acceleration level. Options: none, regular |
adjust_fps_for_interpolation |
boolean |
true |
If true, FPS will be multiplied by the number of interpolated frames |
aspect_ratio |
string |
auto |
Aspect ratio. auto derives from input image. Options: auto, 16:9, 9:16, 1:1 |
enable_prompt_expansion |
boolean |
false |
Whether to expand the prompt using an LLM |
enable_safety_checker |
boolean |
false |
If true, input data will be checked for safety |
end_image_url |
string |
(empty) |
URL of the end image |
frames_per_second |
integer |
16 |
Frames per second |
guidance_scale |
number |
3.5 |
Classifier-free guidance scale for prompt adherence |
guidance_scale_2 |
number |
3.5 |
Guidance scale for the second stage of the model |
image_url |
string |
(empty) |
URL of the input image |
interpolator_model |
string |
film |
Frame interpolation model. Options: none, film, rife |
negative_prompt |
string |
(empty) |
Negative prompt for video generation |
num_frames |
integer |
81 |
Number of frames to generate |
num_inference_steps |
integer |
27 |
Number of inference steps |
num_interpolated_frames |
integer |
1 |
Number of frames to interpolate between each generated frame |
prompt |
string |
(empty) |
The text prompt to guide video generation |
resolution |
string |
720p |
Resolution of the generated video. Options: 480p, 580p, 720p |
seed |
integer |
(empty) |
Random seed for reproducibility |
shift |
number |
5 |
Shift value for the video. Must be between 1.0 and 10.0 |
Output
Examples
Cinematic Image Animation
{
"model": "wan-v2-2-a14b-image-to-video",
"version": "0.0.1",
"input": {
"image_url": "https://example.com/sunset-beach.jpg",
"prompt": "Waves gently crashing on the shore as the sun sets, warm golden light",
"resolution": "720p",
"aspect_ratio": "16:9",
"num_frames": 81,
"guidance_scale": 4.0,
"num_inference_steps": 30,
"interpolator_model": "rife"
}
}
Start-to-End Animation
{
"model": "wan-v2-2-a14b-image-to-video",
"version": "0.0.1",
"input": {
"image_url": "https://example.com/morning-city.jpg",
"end_image_url": "https://example.com/night-city.jpg",
"prompt": "Time-lapse of a city transitioning from day to night",
"resolution": "720p",
"aspect_ratio": "auto",
"num_frames": 81,
"seed": 42
}
}
Related Models
Documentation
1---2name: wan-v2-2-a14b-image-to-video3description: Generate videos from images using Wan v2.2 A14B. Trigger: Use when the user wants to animate an image into a video, or requests 'wan image to video', 'animate image', or 'image to video with wan'.4---56# Wan | v2.2 A14B | Image to Video78Animate a still image into a video guided by a text prompt. Features advanced controls including frame interpolation, dual guidance scales, and multiple resolution options.910## Quick Start1112```bash13curl -X POST https://api.eachlabs.ai/v1/prediction \14 -H "Content-Type: application/json" \15 -H "X-API-Key: $EACHLABS_API_KEY" \16 -d '{17 "model": "wan-v2-2-a14b-image-to-video",18 "version": "0.0.1",19 "input": {20 "image_url": "https://example.com/landscape.jpg",21 "prompt": "Camera slowly pans across the landscape as clouds drift by",22 "resolution": "720p",23 "aspect_ratio": "16:9",24 "num_frames": 8125 }26 }'27```2829## Parameters3031| Parameter | Type | Default | Description |32|-----------|------|---------|-------------|33| `acceleration` | string | `regular` | Acceleration level. Options: `none`, `regular` |34| `adjust_fps_for_interpolation` | boolean | `true` | If true, FPS will be multiplied by the number of interpolated frames |35| `aspect_ratio` | string | `auto` | Aspect ratio. `auto` derives from input image. Options: `auto`, `16:9`, `9:16`, `1:1` |36| `enable_prompt_expansion` | boolean | `false` | Whether to expand the prompt using an LLM |37| `enable_safety_checker` | boolean | `false` | If true, input data will be checked for safety |38| `end_image_url` | string | (empty) | URL of the end image |39| `frames_per_second` | integer | `16` | Frames per second |40| `guidance_scale` | number | `3.5` | Classifier-free guidance scale for prompt adherence |41| `guidance_scale_2` | number | `3.5` | Guidance scale for the second stage of the model |42| `image_url` | string | (empty) | URL of the input image |43| `interpolator_model` | string | `film` | Frame interpolation model. Options: `none`, `film`, `rife` |44| `negative_prompt` | string | (empty) | Negative prompt for video generation |45| `num_frames` | integer | `81` | Number of frames to generate |46| `num_inference_steps` | integer | `27` | Number of inference steps |47| `num_interpolated_frames` | integer | `1` | Number of frames to interpolate between each generated frame |48| `prompt` | string | (empty) | The text prompt to guide video generation |49| `resolution` | string | `720p` | Resolution of the generated video. Options: `480p`, `580p`, `720p` |50| `seed` | integer | (empty) | Random seed for reproducibility |51| `shift` | number | `5` | Shift value for the video. Must be between 1.0 and 10.0 |5253## Output5455- **Type:** video5657## Examples5859### Cinematic Image Animation60```json61{62 "model": "wan-v2-2-a14b-image-to-video",63 "version": "0.0.1",64 "input": {65 "image_url": "https://example.com/sunset-beach.jpg",66 "prompt": "Waves gently crashing on the shore as the sun sets, warm golden light",67 "resolution": "720p",68 "aspect_ratio": "16:9",69 "num_frames": 81,70 "guidance_scale": 4.0,71 "num_inference_steps": 30,72 "interpolator_model": "rife"73 }74}75```7677### Start-to-End Animation78```json79{80 "model": "wan-v2-2-a14b-image-to-video",81 "version": "0.0.1",82 "input": {83 "image_url": "https://example.com/morning-city.jpg",84 "end_image_url": "https://example.com/night-city.jpg",85 "prompt": "Time-lapse of a city transitioning from day to night",86 "resolution": "720p",87 "aspect_ratio": "auto",88 "num_frames": 81,89 "seed": 4290 }91}92```9394## Related Models9596- [Wan v2.2 A14B Image to Video Turbo](../wan-v2-2-a14b-image-to-video-turbo/SKILL.md) - Faster image-to-video generation97- [Wan v2.2 A14B Text to Video Turbo](../wan-v2-2-a14b-text-to-video-turbo/SKILL.md) - Text-to-video generation98- [PixVerse v5 Image to Video](../pixverse-v5-image-to-video/SKILL.md) - Alternative image-to-video model99100## Documentation101102- [EachLabs API Docs](https://docs.eachlabs.ai)
awesome-genmedia