Wan2.2 Day-0 Support in ComfyUI
The WAN team has officially released the open source version of Wan2.2! We are excited to announce the Day-0 native support for Wan2.2 in ComfyUI!
The model consists of high-noise and low-noise expert models, which are divided according to denoising timesteps, resulting in higher-quality video content.
The model is fully open-sourced under the Apache 2.0 license, enabling commercial use.
Highlights of the Wan2.2
MoE Expert Model Architecture: High-noise experts handle overall layout, while low-noise experts refine details.
Cinematic Aesthetic Control: Professional camera language, supporting multi-dimensional visual controls such as lighting, color, and composition.
Large-scale Complex Motion: Smoothly reproduces various complex motions, enhancing controllability and naturalness of movement.
Precise Semantic Adherence: Understands complex scenes, generates multiple objects, and better restores creative intent.
Efficient Compression Technology: Compared to version 2.1, data is greatly upgraded; the 5B version features high-compression VAE and optimized VRAM usage.
Available Models in ComfyUI
Wan2.2-TI2V-5B: Text/Image to video, FP16
Wan2.2-I2V-A14B: Images to video, FP16/FP8
Wan2.2-T2V-A14B: Text to video, FP16/FP8
Get Started
Update ComfyUI or ComfyUI Desktop to the latest version
Go to Workflow → Browse Templates → Video
Select "Wan 2.2 Text to Video", "Wan 2.2 Image to Video", or "Wan 2.2 5B Video Generation"
Download the model as guided by the pop-up
Click and run any templates!
Example Outputs
Master Every Light
Professional Camera Language
Advanced Camera Movement
Capture Every Emotion
Complex Motion, Natural Flow
Unlimited Artistic Expression
Content creation @Yo9oTatara @SyntaxDiffusion @PurzBeats @ComfyUIWiki
Check our documentation for more details: https://docs.comfy.org/tutorials/video/wan/wan2_2
As always, enjoy creating!