Unlock fluid style and motion transfer using WAN 2.1 FUN and Flux, with this innovative ComfyUI workflow. Designed for expressive and dynamic video generation, it enables seamless animation-to-animation transformations while preserving nuanced body movement and artistic styling.

---

🎯 Features

• Style & Motion Fusion – Combine two videos: one as style, the other as motion input.
• Supports WAN 2.1 FUN & Flux – Leverages cutting-edge video modeling architecture for optimal consistency.
• Fully ComfyUI Integrated – All nodes and preprocessing handled directly in ComfyUI, no external tools required.
• Dynamic Video Composition – Frame-by-frame synthesis with control over fps, resolution, and upscaling.
• Supports CFGZeroStar & DiT Models – Allows for advanced model patching and diffusion skipping techniques.

---

💡 Use Cases

• Stylized Animation – Create fluid animated sequences with custom art styles.
• AI Music Videos – Transfer dance or movement from one video into a unique stylized output.
• Video-to-Video Transfer – Ideal for merging poses, expressions, or dynamics across different video clips.
• Motion Testing for Game Design – Rapidly prototype character movements with visual flair.

---

⚙️ How It Works

1. Prepare Your Inputs

   • Choose a motion source video and a style source video.
   • Ensure both are processed through video_cn_preprocess nodes.

2. Load Essential Components

   • Load WAN FUN 1.3B, WAN 2.1 VAE, and umt5_xxl_fp8 CLIP models.
   • Include CLIP Vision, SkipLayerGuidanceDiT, and RIFE VFI nodes.

3. Apply Preprocessing & Alignment

   • Use DWPreprocessor and resolution normalization.
   • Encode visual style via CLIPVisionEncode with img_restyle.

4. Combine Motion & Style

   • Use WanVideoTeaCacheKJ to handle video-to-video transformation.
   • Utilize ModelSamplingSD3 and CFGZeroStar to optimize generation fidelity.

5. Postprocessing & Output

   • Interpolate with RIFE VFI for smooth playback.
   • Export via VHS_VideoCombine with frame rate control, compression, and preview.

---

Credits: cgpixel23