Wan 2.2 (14B) Pose Transfer: Character Animation & Retargeting

his is a high-end Pose-to-Video pipeline designed to animate a static character image using the motion from a reference video. It leverages the state-of-the-art Wan 2.1 (14B) model for cinema-quality generation.

The workflow uses a "One-to-All" pose detection strategy. It extracts the skeleton/movement from a driving video and applies it to your target character image. It also includes automatic background masking via SAM2 and final frame smoothing using GIMM VFI interpolation.

🔥 Key Features:

1. Wan 2.1 (14B) Engine: Uses the powerful wan2.2_animate_14B_bf16 model for high-fidelity motion and texture adherence.
2. Pose Retargeting: Features the PoseDetectionOneToAllAnimation node, which aligns your character to the driving video's pose (using VitPose/DWPose).
3. Smart Masking (SAM2): Integrated Segment Anything Model 2 to intelligently separate the character from the background during generation.
4. Relighting & Style LoRAs: Includes WanAnimate_relight and lightx2v LoRAs to improve lighting consistency and visual quality.
5. Smooth Playback (VFI): Ends with GIMM VFI Interpolation to smooth out the generated frames for a fluid final video.

Links to my patreon and youtube:

https://www.patreon.com/cw/ASF_community?utm_medium=unknown&utm_source=join_link&utm_campaign=creatorshare_creator&utm_content=copyLink

https://www.youtube.com/@ASF_One-Click-PRO