返回工作流

Wan2.2 Animate: Automatic Character Replacement

Upload a video and a character image. Automatically replace the character in the video using the Wan2.2 Animate model.

下载工作流在 Cloud 中试用

Input

The new character, ideally a person with a clear face and full body.

Describe the new character doing what the person in the video does.

A clip of one person. Their motion and the scene are kept; the person is replaced.

How many frames to take from the start of the video. 0 = the whole clip.

Default: 0

Output height (multiple of 16).

Default: 720

Output width (multiple of 16).

Default: 1,280

Advanced settings

What to avoid.

Leave empty for a new result each run. Enter a number to repeat one.

Sampling steps (the lightx2v LoRA is tuned for about 6).

Default: 6

Run

Output
00:00
Example

试试一个示例

关于此工作流

The 'Wan2.2 Animate: Automatic Character Replacement' workflow is designed to seamlessly replace a character in a video with a new character from an uploaded image. This is achieved using the Wan2.2 Animate model, which specializes in character animation and replacement. The workflow begins by loading your video and character image. It then processes these inputs through a series of nodes, including the ComfyUI-WanAnimatePreprocess for initial setup and the ComfyUI-segment-anything-2 for precise segmentation. This ensures that the character replacement is accurate and visually coherent.

Technically, the workflow leverages a combination of video processing nodes such as VHS_LoadVideo and VHS_VideoInfo to extract and manage video data. The b396d3e6-70cc-4a91-81db-6e0399a4edb6 node is crucial for handling complex data transformations required for character animation. By using these nodes in conjunction with the Wan2.2 Animate model, users can achieve high-quality character replacements that are both stable and realistic, especially when the reference image is a clear depiction of a human subject.

常见问答