Seedance 2.5 Reference to Video
Upload reference images to lock in subject and style, add a prompt for motion and mood, and Seedance 2.5 generates video that stays visually consistent.
Seedance 2.5 Reference to Video — How It Works
Upload reference images to guide subject, style, and composition. Add a prompt for motion and mood, and Seedance 2.5 generates video that stays visually consistent across shots.
Everything You Need for Reference-Guided Video Generation
Style-consistent video generation powered by your own reference materials.
Image References
Upload 1-3 reference images to guide the visual style, color palette, subject, and composition of the generated video.
Image-Only References
Veo 3.1 Reference to Video uses image references only. Video uploads are not supported by this model endpoint.
Reference Image Set
Use the first reference image as the strongest subject cue, then add up to two more optional images for style and composition context.
Prompt Control
Add a text prompt to describe the video style, motion, and mood. Combine with references for precise creative control.
Prompt Only
Describe the action, framing, and mood directly in your prompt. Prompt expansion is not submitted for Veo 3.1 Reference to Video.
Full-Resolution Output
Download your generated video in full resolution — ready for social media, presentations, or any professional use.
How It Works
Three simple steps from references to video.
Upload Your References
Upload 1-3 reference images that represent the subject, visual style, and composition you want in the output.
Add a Prompt
Describe the video style, motion, and mood in plain text. The Veo 3.1 request is submitted with your prompt and reference images.
Download the Video
Your style-consistent video is ready in seconds. Download it in full resolution and use it anywhere.
Built for Every Creative Need
From brand consistency to creative projects — reference-guided video fits every workflow.
Brand Video Production
Generate videos that match your brand's visual identity by using existing brand assets as style references.
Style Transfer
Apply the visual style of your reference images to new video content — consistent color grading, lighting, and aesthetic.
Character Animation
Use character or product images as subject references and guide the intended movement through your prompt.
Creative Exploration
Combine multiple reference styles to explore new visual directions and generate unique, unexpected results.
Social Media Content
Create on-brand video content for Instagram, TikTok, and YouTube using your existing visual assets as guides.
Film & Production
Generate style-consistent video sequences for pre-visualization, mood boards, and concept development.
Tips for Better Reference-to-Video Results
How to choose and combine references for the most consistent, high-quality output.
Use References with Consistent Lighting
Reference images with similar lighting conditions produce more cohesive output. Mixing bright outdoor and dark indoor references can create inconsistent results.
Choose References That Share a Visual Style
The AI blends the style of all your references. For the most consistent output, use references that share a similar color palette, contrast level, and visual aesthetic.
Use Focused Reference Images
Use clear, focused images with one main subject or style direction. Describe motion directly in the prompt.
Use the Reference Image Set for Subject Control
Place the most important subject or style in Image 1. Additional images should support the same visual direction rather than introduce conflicting cues.
Combine Up to Three Images
Use up to three aligned images to guide subject, style, and composition. Use the prompt to describe motion and camera behavior.
Write Specific Prompts for Complex Styles
When working with complex or nuanced reference styles, write the key style, motion, and camera details directly in your prompt.
Who Is Reference to Video For?
Anyone who needs style-consistent video output guided by their own visual assets.
Brand & Marketing Teams
Generate videos that match your brand's visual identity by using existing brand assets — photography, style guides, and campaign videos — as references.
Animators & Motion Designers
Use image references plus detailed prompts to guide motion style and character animation concepts.
Film & Video Producers
Generate style-consistent pre-visualization footage, mood board videos, and concept sequences using reference materials from the production.
Social Media Agencies
Create on-brand video content for clients by using their existing visual assets as style references — consistent output at scale.
Game Developers
Generate cinematic sequences and environment videos that match the visual style of your game using concept art and image references as guides.
Creative Directors
Explore and validate visual directions quickly by generating reference-guided video concepts before committing to full production.
Why Choose Veo Reference to Video?
Built for style consistency, quality, and creative control.
Style Consistency
Use your own reference materials to ensure the generated video matches your exact visual style and brand identity.
Results in Seconds
Most generations complete in under 60 seconds. No waiting, no queues, no batch processing delays.
Privacy First
Your uploaded files are processed securely and never used to train AI models or shared with third parties.
Pay Only for What You Use
Credit-based pricing means you only pay for the videos you actually generate — no monthly minimums.
Frequently Asked Questions
Everything you need to know about AI reference-to-video generation.
What types of reference files can I upload?
For Veo 3.1 Reference to Video, upload 1-3 reference images (PNG, JPG, WEBP). This model endpoint does not support reference video uploads.
What is Image 1 used for?
Image 1 is the strongest subject cue for the generated video. Additional images can reinforce the same subject, style, or composition.
Do I need to upload videos?
No. Veo 3.1 Reference to Video requires images only: minimum 1, maximum 3 reference images. Video uploads are not supported on this model endpoint.
How does the prompt work with references?
The prompt describes the video style, motion, and mood in plain text. It works together with your references — the references guide the visual style while the prompt guides the content and action.
What is prompt expansion?
Prompt expansion is not submitted for Veo 3.1 Reference to Video. Write the important action, style, camera, and mood details directly in your prompt.
How long will the generated video be?
Video length depends on the model and settings selected. Available duration options are shown before you generate.
How many credits does a generation cost?
Credit cost depends on the model, resolution, and duration selected. The required credits are shown before you generate, so there are no surprises.
Is my data private?
Yes. Your uploaded files and generated videos are processed securely and are not used to train AI models or shared with third parties. Files are automatically deleted after processing.
Can I use generated videos commercially?
Yes. Videos generated using Seedance 2.5 are yours to use for commercial purposes. Check the terms of service for full details on usage rights.
How many reference files should I upload for best results?
For most use cases, 1–3 focused reference images produce the best balance of style consistency and generation quality. Too many conflicting references can dilute the style signal.
Can I use screenshots from a film or TV show as references?
Technically yes, but be mindful of copyright. Using copyrighted material as a reference to generate commercial content may raise legal issues. For commercial projects, use your own assets or royalty-free reference materials.
What is the difference between reference-to-video and text-to-video?
Text-to-video generates content purely from a written description, giving the AI full creative freedom. Reference-to-video uses your uploaded images to constrain the visual style and subject direction, producing output that is consistent with your specific reference materials.