Wan Animate is the character-performance specialist built on Alibaba’s Wan 2.2 architecture. Instead of imagining motion from a prompt, it borrows motion from footage you provide — a dancer, an actor, a phone clip of yourself — and applies that performance to a character you choose. It’s the difference between describing choreography and capturing it.
Flux 3 AI hosts both Animate modes. Move is motion transfer: your character image performs the movements from the driving video in a newly generated scene. Replace goes the other way, swapping the person inside the original footage for your character while the background and action carry on unchanged. Both take an image plus a video input, support a reference image, and include an audio setting for the source clip.
Results render at 5 or 10 seconds, 720p or 1080p, in 16:9 or 9:16. For best output, use driving footage with one clearly visible person and a character image in a comparable pose — clean inputs are what separate an uncanny swap from a convincing one.
See what's possible on Flux 3 AI
Sample generations from the platform — write your own prompt to see what Wan Animate does with it.
Wan Animate vs similar AI video models
Specs from the live catalog — every model below is available on the same account.
| Spec | Wan Animate | Wan 2.2 Turbo | Wan 2.7 |
|---|---|---|---|
| Max clip length | 10s | 8s | 10s |
| Durations | 5s, 10s | 4s, 6s, 8s | 5s, 10s |
| Resolutions | 720p · 1080p | 480p · 720p | 720p · 1080p |
| Aspect ratios | 2 | 2 | 3 |
| Audio | |||
| Image to video | |||
| Reference images | up to 1 | up to 1 | up to 9 |
| Keyframes | |||
| Negative prompt |
Capabilities reflect what each model supports on Flux 3 AI today.
What makes Wan Animate stand out
Move: motion transfer
Your character inherits the exact performance from a driving video — gestures, dance steps, timing and all.
Replace: character swap
Keeps the original scene and action intact while substituting the on-screen person with the character you upload.
Real footage as director
Motion comes from actual video rather than a text description, giving performances a physicality prompts rarely hit.
Audio handling built in
An audio setting on both modes controls how the driving clip’s sound carries into the finished render.
1080p deliverables
Both Animate modes render 5- or 10-second clips at up to 1080p in widescreen or vertical framing.
Popular use cases
Why creators run Wan Animate on Flux 3 AI
- Try the same driving clip on Wan Animate and Kling Motion Control, then keep the cleaner transfer.
- Finished 1080p performances export watermark-free from your unified generation library.
- No dedicated Wan plan required — one shared credit balance covers Animate and everything else.
Wan Animate specs & capabilities
Output
- Clip duration5, 10 seconds
- Resolutions720p · 1080p
- Aspect ratios16:9 · 9:16
- Generation modesMotion Transfer · Character Replacement
Controls
- Image to video (start frame)
- Reference images
- First & last frame keyframes
- Audio generation
- Negative prompt
- Video input / editing
- Extend video
Available versions on Flux 3 AI
Every Wan Animate endpoint below is selectable in the Create studio.
- Wan 2.2 Animate MoveWan 2.2 Animate — motion transfer
- Wan 2.2 Animate ReplaceWan 2.2 Animate — character replace
Wan Animate prompt examples
Copy one as a starting point, or send it straight to the Create studio.
“Transfer the choreography from this dance clip onto the uploaded knight character, keeping every spin and footstep, staged in a torch-lit castle hall”
Move mode: motion transfer into a newly generated scene
“Swap the presenter in this cooking video for the uploaded chef character, preserving the kitchen, the plating, and the original camera movement”
Replace mode: character swap while the filmed scene stays intact
“Map the skateboarder’s kickflip onto a cartoon fox character, exaggerated squash-and-stretch on the landing, golden-hour skatepark in the background”
Tests real physical timing carried into a stylized character
Create with Wan Animate in three steps
Describe your idea
Write a prompt — the more specific the scene, motion, and style, the better Wan Animate performs.
Pick Wan Animate & settings
Choose duration, resolution, and aspect ratio. The studio shows the exact credit cost before you generate.
Generate & download
Wan Animate renders in the cloud — track progress in your library, then download watermark-free or share with a link.
Wan Animate questions, answered
What is Wan Animate?
Wan Animate is a pair of Wan 2.2-based tools on Flux 3 AI for character performance. The Move mode transfers motion from a driving video onto a character image; the Replace mode swaps the person in existing footage for your character while preserving the scene.
What’s the difference between Animate Move and Animate Replace?
Move generates a new video of your character performing the motion captured from the driving clip. Replace edits the driving clip itself, keeping its background and action but substituting the visible person with your uploaded character. Same inputs, opposite outputs.
What inputs does Wan Animate need?
Both modes require two things: a character image and a source video that supplies the motion. An optional reference image can further anchor the character’s appearance, and an audio setting determines how the source clip’s sound is handled.
How long and what quality are Wan Animate videos?
Each render is 5 or 10 seconds at 720p or 1080p, in 16:9 widescreen or 9:16 vertical. Clips download from your Flux 3 AI library at full quality with no watermark.
Can I put myself into an existing video with Wan Animate?
Yes — that’s exactly what Replace mode does. Upload the target footage and a clear photo of yourself, and the model swaps you in while the rest of the scene continues as filmed. A front-facing, well-lit photo produces the most convincing result.
What footage works best as the driving video?
Clips with a single, fully visible person, steady framing, and distinct motion transfer most cleanly. Crowded scenes, heavy occlusion, or rapid camera cuts make the motion harder to track and lower the quality of both Move and Replace outputs.