Wan2.2 Animate GGUF - Video Animation Workflow

Overview

This ComfyUI workflow enables high-quality video animation and character motion transfer using the Wan2.2-Animate-14B model in GGUF format. It's specifically designed for creating animated videos from reference images and motion source videos.

Key Features

🚀 GGUF Model Optimization

• Uses GGUF format for efficient memory usage and faster loading

• Compatible with various hardware configurations

• Includes separate GGUF loaders for model, CLIP, and VAE components

🎭 Dual Operation Modes

• Character Replacement Mode: Replace characters in existing videos while preserving background

• Motion Transfer Mode: Apply character poses to new scenes and environments

🛠️ Advanced Preprocessing

• Interactive point-based segmentation using SAM2

• Automatic pose detection with DWPreprocessor

• Facial feature extraction for better character preservation

• Smart video scaling and frame management

Workflow Structure

Step 1: Model Loading

• Loads Wan2.2-Animate-14B GGUF model

• Configures CLIP text encoder and VAE decoder

• Applies optional LoRA enhancements for improved results

Step 2: Input Setup

• Reference image upload for character appearance

• Source video for motion capture

• Positive/Negative prompt configuration

Step 3: Video Preprocessing

• Extracts frames, audio, and FPS from source video

• Resizes video to optimal dimensions (must be multiples of 16)

• Generates pose and facial reference data

Step 4: Character Masking

• Interactive Points Editor for precise character selection

• SAM2 segmentation with positive/negative point guidance

• Mask refinement with GrowMask and BlockifyMask nodes

Step 5: Animation Generation

• Dual KSampler setup for flexible video generation

• WanAnimateToVideo nodes handle core animation logic

• Support for video length extension through batch processing

Step 6: Video Output

• Recombines generated frames with original audio

• Maintains original FPS for seamless playback

• Multiple output options with SaveVideo nodes

Technical Requirements

Hardware

• Compatible with various GPU/CPU configurations thanks to GGUF format

• Lower VRAM requirements compared to standard model formats

• Recommended: 8GB+ RAM for optimal performance

Software

• ComfyUI with required custom nodes:

  • ComfyUI-segment-anything-2 (SAM2)

  • comfyui-controlnet-aux (preprocessors)

  • comfyui-kjnodes (utility nodes)

  • GGUF loader nodes

Usage Instructions

1. Load Models: Ensure all GGUF model files are in correct directories

2. Set Dimensions: Configure width/height as multiples of 16 (e.g., 640x640)

3. Input Media: Upload reference image and source video

4. Mask Creation: Use Points Editor to mark character areas (Shift+click for positive points)

5. Configure Prompts: Set positive and negative text prompts

6. Execute: Run the workflow and monitor progress through preview nodes

Ideal For

• Character animation from still images

• Motion transfer between videos

• Video style transfer with character preservation

• Content creation for short films and social media

This workflow represents a sophisticated pipeline for video animation that balances quality with computational efficiency through the use of GGUF model format.