Video character replacement icon

Video Replace – Wan 2.2 Animate

Provide a reference video and a character image. The model replicates expressions, body motion, and gestures onto your character while preserving lighting and scene context.

Replace performer with your character

Ideal for character animation, face-forward creative edits, and replacing actors while keeping the original environment. Reference video duration must be at most 8 seconds. Output width and height are each between 256 and 852 pixels.

Reference video

Click or drop a reference video (MP4, WEBM, MOV).

Max 8 seconds. Large files may take longer to upload.

Character image

Upload the character / target image (JPG, PNG, WEBP).

Max 10 MB.

Optional prompt

Output size

Each side must be between 256 and 852 px (inclusive).

Result

Output video will appear here.

Model highlights

  • 14B parameters, MoE architecture — strong quality with efficient inference.
  • Skeleton-driven motion and implicit facial reenactment from reference frames.
  • Relighting-style adaptation for compositing with the original scene.
  • Works for humans, illustrated, or stylized characters.

What this tool does

Animation: Drive a static character with motion and expressions copied from a performer in your reference clip.

Replacement: Swap the on-screen performer for your character image while keeping lighting, color tone, and background consistency where the model allows.

Common uses: brand mascots from one image, motion transfer to digital avatars, actor replacement for creative edits, and short clips with a consistent character identity.

Frequently asked questions

What are the video limits?

Reference videos should be at most 8 seconds. Use a short, clear clip with a visible performer for best motion transfer.

Which output resolutions are allowed?

Width and height must each be between 256 and 852 pixels. The default is 768×768; you can switch presets or type custom values within that range.

Do I need a prompt?

No. Prompt is optional and can be used for extra guidance when the API supports it.