Combine visual decisions
Use separate images for identity, wardrobe, products, locations, composition, or style.
Direct a video with multiple visual references
References to Video AI Generator
Keep identity, products, and style connected
Use multiple reference images to direct identity, wardrobe, products, environments, composition, and visual style in one Seedance 2.5 video. Assign every reference a clear role, then describe the action, camera, continuity, and sound.
Built-in generator
Create a reference-guided video with Seedance 2.5
Seedance 2.5 supports reference-aware video direction with two or more compatible images. Upload a focused reference set, explain what each image controls, and review the exact cost before generating.
Use the smallest useful reference set and assign one clear responsibility to every image. Paid credits are required, and the exact total is shown before generation.
Why use this workflow
More control from the first frame to the final beat
Protect important details
State which face, material, logo, silhouette, palette, and environment details must remain stable.
Direct one coherent shot
Turn several sources into one ordered action, camera path, atmosphere, and final frame.
How it works
Create with References to Video in three steps
- 1
Choose compatible references
Upload only the images needed to define identity, product, wardrobe, location, composition, or style.
- 2
Assign roles and direct the shot
Name what each image controls, resolve conflicts, and describe the action, camera, protected details, and sound.
- 3
Review settings and generate
Choose duration, aspect ratio, resolution, format, and audio, then confirm the displayed credit cost.
Model choice
Start with Seedance 2.5, switch when the brief requires it
Seedance 2.5 is the default for longer 4–30 second sequences, richer continuity direction, MP4 or MOV output, and optional native audio. Seedance 2.0 remains available for 5–15 second clips and resolution options up to 4K.
Frequently asked questions
References to Video AI Generator FAQ
What is References to Video?+
References to Video creates a new video from several visual sources. Each image can define a different responsibility, such as identity, wardrobe, product geometry, environment, composition, lighting, or style.
How many reference images should I use?+
Begin with the smallest set that communicates the required visual facts. More images are useful only when they add a distinct role without introducing conflicting information.
When should I use Image to Video instead?+
Use Image to Video when one source image already defines the subject and composition. Use References to Video when the final shot must combine information distributed across multiple assets.
How can I reduce identity or product drift?+
Use clear sources, assign one image as the authority for each protected detail, resolve conflicts in writing, and keep the requested action compatible with the references.