← All models
Restyle a video’s scene, lighting, and visual style from edited keyframes while preserving the source subjects’ identity, expressions, gaze, and motion. Developed by Eyeline Labs and Netflix researchers.
Example
Details
- Accepts
- image + video
Pricing
- Price
- 25 cr / second
Prices in credits (1 credit = $0.01).
Data schema
Input
| Field | Type | Description |
|---|---|---|
| seed | — | Random seed for reproducibility. The same seed is used for every clip of a multi-clip generation. |
| promptrequired | string | The prompt describing the restyled video to generate. |
| image_urlrequired | string | URL of the restyled first frame. The video is regenerated in this frame's identity/style. |
| keyframes | array | Optional restyled keyframes pinned at specific output frame indices to anchor identity over long videos. |
| video_urlrequired | string | URL of the source video whose motion and framing drive the generation. |
| num_frames | integer | Maximum number of output frames. Capped at the source video's frame count. |
| resolution | stringenum: 480p, 720p | Output resolution tier. Orientation follows the source video (720p is 1280x720 or 720x1280; 480p is 832x480 or 480x832). |
| vace_scale | number | VACE conditioning scale. Controls how strongly the source video's structure constrains the generation. |
| guidance_scale | number | Classifier-free guidance scale. Controls prompt adherence versus creativity. |
| num_inference_steps | integer | Number of diffusion sampling steps per clip. Higher improves quality but is slower. |
| segmentation_prompt | string | Text prompt for SAM-3 subject segmentation. The segmented subject's pixels condition the generation; the rest of the frame is regenerated freely. |
| enable_safety_checker | boolean | Enable safety checking of inputs and outputs. |
Output
| Field | Type | Description |
|---|---|---|
| seed | integer | The seed used for generation. |
| video | — | The generated video file. |