wan2.2-animate-move vs animate-mix, and Sume's motion transfer
animate-move drives a still character with a reference video; animate-mix swaps a character into a video. Sume's Genjutsu row covers the move-style case.

The two Wan 2.2 models solve opposite problems. wan2.2-animate-move takes a static character image and moves it with the motion of a reference video, keeping the background unchanged. wan2.2-animate-mix takes a video and replaces the character in it with one from a source image. Sume has a single motion-transfer row, Genjutsu, and it matches the move case only.
Alibaba's side is from its Model Studio video generation page, read 2026-10-01. Sume's side is from the Video Router docs.
What is the difference between move and mix?
Both are listed for 720P output and 2-30 s. Alibaba's page describes both as having a pro mode (wan-pro) and a standard mode (wan-std); the page gives the pro mode as closer to live-action footage. The page does not list a price in the section read here, so none is claimed.
| Model | Input | What changes | Output |
|---|---|---|---|
wan2.2-animate-move | Static character image + motion reference video | The character moves; the background remains unchanged | 720P, 2-30 s |
wan2.2-animate-mix | A video + a source image of the new character | The character in the video is replaced | 720P, 2-30 s |
Which one does Sume's motion transfer match?
The move case. In the Video Router catalog the Genjutsu row is described as Motion Transfer only: a video_url plus 1-8 reference_image_urls, with no text-only generation. The video supplies the motion and the images supply the character. Sume's docs do not describe a character-replacement mode for it, so treat animate-mix as having no equivalent there.
If the goal is to keep a person's performance and swap who appears, that is the mix task. Sume's other video_url row, Gemini Omni Flash 1.1, is an edit row driven by a text prompt; see edit a video with a prompt for what it accepts.
What are Sume's limits for this row?
Per the Video Router docs, the row serves 480p and 720p, 4-30 s, and the duration is the input video's length, rounded up. It sends no aspect_ratio, generate_audio, bitrate_mode or audio references. It is also listed only when its provider is configured, so check GET /v1/video-router/models before building on it.
Alibaba's 2 s minimum is lower than Sume's 4 s, so a 2-3 s motion clip that works there needs to be 4 s or longer here. Inspect the source clip first.
What should I do?
Pick by task. To animate a still with a reference performance, send the clip and the character images to the Genjutsu row. To swap a character inside existing footage, use a tool built for replacement; Sume's catalog does not list one. Read capabilities from the catalog rather than assuming an envelope.
Sources
Related posts
More in Models
- wan2.7-videoedit camera replication vs Sume's video_url edit
Alibaba names wan2.7-videoedit for effect and camera-movement replication. On Sume, video edit is the video_url field on the Gemini Omni Flash 1.1 id.
- Wan 3.0 Prime vs Wan 3.0: one wan-3.0 id on Sume
QwenCloud and Runway list Wan 3.0 and a Prime variant. Sume's catalog lists one id, wan-3.0, with 2-30 s clips. Check the catalog for limits.
- Wan 3.0 is public after invite-only: check Sume's wan-3.0 id
QwenCloud says Wan 3.0 went from invite-only on August 6 to public. On Sume, wan-3.0 is a catalog id; check the live catalog for limits.
- Which Sume video models accept an audio reference?
Seven Sume video models take reference audio: the four Seedance 2.x ids, Wan 3.0 and the two MiniMax H3 models. Kling, Grok and Gemini Omni Flash do not.
Written by Sume