Combine Images, Video & Audio
Use different media for different parts of your creative direction. Images can define characters and visual style, video can guide motion and camera behavior, and audio can influence timing, rhythm, or atmosphere.
Multi-Reference AI Video Generation
SeedGen AI Reference to Video Generator
Combine images, video clips, and audio references to guide your AI video. Keep characters, visual style, motion, and scene direction more consistent while controlling the result with your prompt.
Runs in the browser · no install · new accounts start free
What Is SeedGen AI Reference to Video?
SeedGen AI Reference to Video lets you create videos using multiple images, video clips, and audio files as creative references. Use it when one image alone cannot fully define the characters, visual style, motion, timing, or sound you want.
Upload the reference assets that matter to your shot, then use a prompt to tell SeedGen how they should work together. Different references can help guide subjects, environments, composition, movement, pacing, and audio direction within the same generation.
Supported limits depend on the AI video model you select. For example, Seedance 2.5 supports up to 50 multimodal references, giving you more room to build complex scenes and maintain creative continuity. See the 50-reference guide for a deeper workflow walkthrough.
If your shot is already defined by image inputs and you mainly want to add motion, use SeedGen AI Image to Video instead.
SeedGen AI Reference to Video Features
Combine images, video clips, and audio references to give the model more context than a single image can provide. Guide characters, motion, visual style, timing, and scene direction with the assets that matter to your shot.
Up to 50 Multimodal References
On supported models such as Seedance 2.5, add up to 50 reference assets in one generation. Build more complex shots with multiple characters, environments, visual details, motion examples, and audio cues without relying on a single source.
Reference Specific Assets in Your Prompt
Use references such as @Image1, @Video1, or @Audio1 to tell SeedGen exactly which uploaded asset should guide a character, action, scene, or creative detail.
Keep Creative Direction Connected
Use multiple references together to maintain stronger continuity across characters, environments, style, motion, and timing. This is especially useful when one image cannot fully describe the shot you want.
How to Use SeedGen AI Reference to Video
Combine image, video, and audio references to guide your generation in four simple steps.
Upload Your Reference Media
Choose Multi Reference, then upload the images, video clips, and audio files that should guide your result. Use only the references that add useful information about the subject, scene, motion, style, or sound.
Reference Assets in Your Prompt
Describe the video you want and use markers such as @Image1, @Video1, or @Audio1 to tell SeedGen which uploaded asset should guide each part of the generation.
Adjust Your Video Settings
Choose the duration, resolution, and aspect ratio for your video. Check the reference limits shown in the generator, including supported media duration and file requirements.
Generate and Preview Your Video
Click Generate Video to start the generation. Preview the result in the output panel, then refine your prompt, references, or settings and generate another version if needed.
SeedGen Reference to Video Examples
Explore two AI videos created with multiple reference images. Open each example to see the prompt, settings, and reference assets used to generate the result.
Friday-night football stills
4s · 1080p · 16:9Reference input
View prompt and notes
Use @Image1 for the football and @Image2 for the empty Friday-night field. Keep the ball on the wet turf under stadium lights. Slow push toward the ball, light mist, no players, no crowd, no logos.
Seedance 2.5 · two stills · 4s · 1080p · 16:9 · Smart prompt on
Reference inputs are the football still and the empty field still. This clip was generated on this page, not reused from the site library.
Diner shake on the booth table
4s · 1080p · 16:9Reference input
View prompt and notes
Use @Image1 for the stainless-steel chocolate malt milkshake and @Image2 for the empty red vinyl diner booth. Place the shake on the booth table by the night window. Slow push-in, pink neon glow, condensation on the cup, no people, no faces, no logos, no extra food.
Seedance 2.5 · two stills · 4s · 1080p · 16:9 · Smart prompt on
Reference inputs are the malt shake still and the empty diner booth still. This clip was generated on this page, not reused from the site library.
SeedGen Reference to Video Use Cases
Use multiple references when one image can’t fully define the people, products, environments, style, motion, or sound you want in the final video.
Multiple Characters & Subjects
Use separate references for different people, characters, objects, or products so each important subject has a clear visual source.
Products in Specific Environments
Combine a product image with a location or lifestyle reference to keep both the product and setting aligned with your creative direction.
Style & Visual Direction
Use multiple images to guide color palette, lighting, texture, wardrobe, composition, and environment when one reference can’t capture the full look.
Motion & Audio Guidance
Add video or audio references when movement, camera behavior, timing, rhythm, dialogue, or atmosphere are important to the result.
SeedGen AI Reference to Video FAQ
Quick answers about reference files, supported formats, limits, video settings, credits, audio, and commercial use.
SeedGen AI Reference to Video lets you create AI videos using multiple images, video clips, and audio files as references. Upload the assets you want to follow, describe how they should work together, and generate your video directly in the browser.
SeedGen supports common reference formats including JPG, PNG, WebP, BMP, TIFF, and GIF for images, MP4 and MOV for video, and MP3 and WAV for audio. Available formats may vary by model.
Reference limits depend on the AI video model you select. Seedance 2.5 supports up to 50 reference assets in one generation, including images, video, and audio.
Yes. On the current Seedance 2.5 workflow, each video or audio clip must be at least 2 seconds long. Total video references can be up to 30 seconds, and total audio references can also be up to 30 seconds.
Available settings depend on the selected model. Supported options can include 4–30 second videos, resolutions from 480p to 1080p, and common landscape, square, portrait, and vertical aspect ratios.
Image to Video is best when image inputs already define the main visual direction. Reference to Video is designed for more complex generations that need multiple images, video clips, or audio references to guide characters, environments, motion, style, timing, or sound.
Yes. New SeedGen users can start with available free credits. The required credit amount is shown before generation, and additional credits can be purchased when needed.
Yes. Commercial use is supported subject to the SeedGen Terms and any applicable model-specific restrictions. You must have the necessary rights to the images, video, audio, likenesses, and other content you upload or reference.
Credit packs and per-second rates are on SeedGen pricing.
Turn Your References Into a Video
Combine your images, video clips, and audio references to shape the characters, motion, style, and sound you want.
Create Your Video