alibaba/wan-2.2/animate

Open and Advanced Large-Scale Video Generative Models.

VIDEO-TO-VIDEO
video-to-video

Open and Advanced Large-Scale Video Generative Models.

Wan2.2-Animate

What is Wan2.2-Animate?

Wan2.2-Animate is an optimized, unified character animation/replacement model that maps static inputs to targeted actions, replicating holistic body movement and facial expressions with high temporal stability and natural realism.

How to Use (Wan2.2-Animate)

  1. image – Upload a clear character photo. (Recommend using PNG and JPG formats, and avoid WEBP)
  2. video – Upload the motion clip (pose/expressions come from here).
  3. prompt (optional) – Brief rules, e.g., “preserve outfit; natural expression; no background changes.”
  4. mode – Choose replace or animate.
  5. resolution – Pick 480p or 720p.
  6. Generate – Wait a moment for results.
  7. Review & iterate – Fix a seed to reproduce; change seed for A/B variants.

Tips

  • Match composition & pose: Keep the reference image and target video aligned in composition, camera position, and human body pose to reduce failure rates.
  • Keep aspect ratios the same: Use the same aspect ratio for the input image and video for best results.

Specifications in Depth

Overview:

Model Provider:QWEN
Model Type:video-to-video
Deployment:Inferencing API; Playground
Pricing:$0.0400/second

Key Specs:

Size Cap:up to width × height (user-configurable)
LoRA Support:No
Seed Options:N/A

Create Your Next Masterpiece

Inizia con Oltre 300 Modelli,

Solo su Atlas Cloud.