ワークフローに戻る

3.2 Advanced: Control Video Generation

Learn how to control a video with Pose Estimation and Wan Animate.

ワークフローをダウンロードCloudで試す

Input

The character to animate; works best fully in frame.

Describe the character and the scene.

The clip whose movement the character copies; the pose is estimated automatically.

Output height; 1280×720 for widescreen.

Frames to generate at the pose video's frame rate; 73 is about 3 s at 24 fps. Use 4n+1 values.

Output width; 1280×720 for widescreen.

Advanced settings

1 with the speed LoRA.

Default: 1

What to avoid.

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

6 with the lightx2v speed LoRA.

Default: 6

Run

Output
00:00
Example

例を試す

このワークフローについて

This advanced ComfyUI workflow is designed to empower users with the ability to control video generation using pose estimation techniques and the Wan Animate model. At its core, the workflow leverages the DWPose Estimator to extract pose information from an input video, creating a 'control video' that serves as a blueprint for generating new video content. This control video is then used in conjunction with the Wan Animate model to produce videos that follow the poses and movements captured in the original footage.

The workflow involves several key nodes, including VAEDecode for decoding latent representations, CreateVideo for assembling frames into a video, and SaveVideo for outputting the final product. Additionally, the comfyui_controlnet_aux custom node is utilized to enhance control over the video generation process, ensuring that the output aligns closely with the user's vision.

This workflow is particularly useful for animators and content creators who wish to produce videos that maintain consistent motion and style, guided by specific pose references.

よくある質問