Asset reference
Upload reference assets to guide the AI generation process for more precise creative control. Both single-asset and multi-asset combination references are supported.
Single-asset reference#
- Image reference — AI preserves the style, tone, and composition of the image and generates a dynamic video
- Video reference — AI references the camera movement, pacing, and visual style
Combination reference#
Multiple assets can be freely combined, and AI will comprehensively interpret all of them for generation:
| Combination | Use case |
|---|---|
| Image + Audio | Use an image to set the style and audio to set the rhythm and atmosphere |
| Image + Video | Use an image to set the visual style and a video to set the camera movement |
| Video + Audio | Use a video as visual reference and audio as background music |
| Image + Video + Audio | Full-spectrum reference for precise control over the generated output |
First frame / first-and-last frame mode#
Upload an image as the first frame of the video, and AI generates the subsequent frames. You can also upload both a first frame and a last frame image, and AI will automatically generate a smooth transition video between them. Ideal for product showcases, scene transitions, and other scenarios requiring precise visual control.
Frame mode vs. reference images#
Face references#
On the Seedance series, just upload a face reference image directly — face compliance is applied automatically on submit, with no character or asset registration required. Reuse the same reference across shots to keep a character consistent.