AI Tools / Video

Video to Video AI

Guide a new clip from reference video—not still first/last frames. Upload 1–3 short reference clips, optionally add images or audio, describe what to keep or change, and generate with Seedance 2.0.

Try Video to Video See pricing

Seedance 2.0 only · reference videos required (1–3 clips)
Optional: up to 9 reference images + up to 3 audio clips
Not Image to Video — that tool uses first/last frame stills

Reference videos: 480–720px short side, ≤15s combined

Optional image_urls (not frame roles)

No real people in reference footage (UI rule)

Prompt up to 4,000 characters

Models

Which model powers Video to Video?

Dashboard → Video to Video loads models from the catalog; today only Seedance 2.0 is supported for this workflow.

Seedance 2.0 — duration 4–15 seconds (default 5); aspects 16:9, 9:16, 1:1, 4:3, 3:4, 21:9; quality 480p, 720p, or 1080p
Optional sound: With Sound / Not With Sound
Advanced: With Last Frame / Not With Last Frame toggle (Seedance output option—not the same as Image to Video frame uploads)
VEO 3.1 is not available on this tool—use Text to Video or Image to Video for VEO

Inputs

What do you submit?

Video to Video is reference-clip driven. It is not first/last frame animation (Image to Video) and not a 1–9 still board without video (Multi-images to Video).

Reference videos (required): 1–3 clips. Combined length ≤15 seconds. Each clip short side must be 480–720 pixels (portrait 480p–720p supported). UI: “Do not use footage that contains real people.”
Reference images (optional): up to 9 images sent as image_urls—style, character, or product guides—not first frame / last frame roles.
Reference audio (optional): up to 3 clips, combined ≤15 seconds—for rhythm or mood hints.
Prompt (required): up to 4,000 characters (~500 recommended) describing what to keep vs change from references.
Output settings: duration, aspect ratio, quality, sound, optional public gallery.
At least one valid reference video is required before Generate succeeds.

Reference media

What makes good reference videos, images, and audio?

Short, clear motion clips

Use 1–3 tight clips under 15 seconds total. Simple camera paths and readable motion are easier to follow than busy long takes. Trim before upload.

480–720px short side

Scale reference video so the shorter dimension is within 480–720px. Portrait 480p–720p is supported. Clips outside this band are rejected with a validation error in the UI.

Content rules

Do not upload reference footage containing real people. Use synthetic, animated, or licensed stock without recognizable individuals.

Prompting

Prompt tips: keep vs change

Keep camera path, change environment

Example: “Keep the dolly path and timing from the reference. Replace the background with a futuristic city at dusk; match lighting on the subject.”

Keep subject, change style

Example: “Preserve character silhouette and walk cycle from reference. Re-render as hand-painted animation with soft brush strokes.”

Optional audio + images

When you add reference audio, describe how it should influence pacing. For optional stills, name roles: “Outfit colors from reference image 1; environment tone from image 2.”

When to use this tool

Video to Video vs other tools

Choose this when motion already exists in reference clips and you want a new take—not when you only have one or two still frames.

Choose Video to Video when: you have short reference clips whose motion, timing, or camera path should guide the output.
Choose Image to Video when: you have first frame (required) and optional last frame stills—not video clips.
Choose Multi-images to Video when: you have 1–9 still guides but no reference video and no defined end frame.
Choose Text to Video when: you have no reference media and can describe the shot entirely in words.
Optional images here supplement video—they are not a substitute for the required reference clips.

Prompt blueprint

Reference-guided prompt template

Keep from reference: camera path, timing, subject motion, or beat structure
Change: environment, style, wardrobe, color grade, or props
Optional images: name how each still should influence the result
Optional audio: describe pacing or mood alignment
Avoid: contradicting the reference motion in the same prompt
Not for: first/last frame still pairs — use Image to Video instead

Examples

Example Video to Video outputs

Swap in your best reference-guided clips for search trust.

Environment swap

Reference motion preserved; scene replaced via prompt.

Style remix

Animated re-render of live-action reference move.

Vertical hook

9:16 reference timing with new visual treatment.

Product motion

Reference product move with new studio look.

FAQ

Is Video to Video the same as Image to Video?

No. Video to Video requires 1–3 reference video clips. Image to Video uses a required first frame still and optional last frame still—no reference video uploads.

How many reference videos can I upload?

1–3 clips. Combined duration must be at most 15 seconds. Each clip short side must be between 480 and 720 pixels (portrait 480p–720p supported).

Can I upload only images without video?

No. At least one reference video is required. Optional images (up to 9) supplement the clip—they use image_urls, not first/last frame roles.

Which model is supported?

Seedance 2.0 only on this tool today.

Why was my reference video rejected?

Check clip count (1–3), total duration (≤15s), short side (480–720px), and content rules—reference footage must not contain real people.

Can I add audio references?

Yes—optional, up to 3 audio clips with combined length ≤15 seconds.

How long can my prompt be?

Up to 4,000 characters; around 500 is enough for most shots.

Can I use outputs commercially?

Commercial usage depends on your plan and terms. Review pricing and policies before shipping at scale.

Reshoot motion from reference clips

Upload short references, describe keep vs change, and generate with Seedance 2.0.

Try Video to Video View pricing