Reference: ComfyUI official Wan2.2 tutorial

1. Download the Models

• Enter the ComfyUI models directory

1

cd models

• Download the repository

1

GIT_LFS_SKIP_SMUDGE=1 git clone https://www.modelscope.cn/Comfy-Org/Wan_2.2_ComfyUI_Repackaged.git

• Download the models

1
2

cd Wan_2.2_ComfyUI_Repackaged
git lfs pull&

2. 5B Hybrid Edition (Text-to-Video + Image-to-Video)

2.1 Download the Workflow

1

wget https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/video_wan2_2_5B_ti2v.json

Or click Workflow → Browse Templates → Video → Wan2.2 5B video generation in turn to load the workflow.

2.2 Generate the Video

• Load Diffusion Model → Wan_2.2_ComfyUI_Repackaged/split_files/diffusion_models/wan2.2_ti2v_5B_fp16.safetensors
• Load CLIP → Wan_2.2_ComfyUI_Repackaged/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors
• Load VAE → Wan_2.2_ComfyUI_Repackaged/split_files/vae/wan2.2_vae.safetensors
• CLIP Text Encoder → fill in the positive / negative prompt
• Wan22ImageToVideoLatent → set width / height / length (number of frames)
• Image-to-video: press Ctrl+B to enable Load Image, then upload an image
• Click Run, or press Cmd+Enter / Ctrl+Enter

3. 14B Text-to-Video (T2V)

3.1 Download the Workflow

1

wget https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/video_wan2_2_14B_t2v.json

Then click Menu -> File -> Open to import the workflow.

Or click Templates -> search for Wan2.2 14B T2V in turn to load the workflow.

3.2 Generate the Video

• The first Load Diffusion Model → Wan_2.2_ComfyUI_Repackaged/split_files/diffusion_models/wan2.2_t2v_high_noise_14B_fp8_scaled.safetensors
• The second Load Diffusion Model → Wan_2.2_ComfyUI_Repackaged/split_files/diffusion_models/wan2.2_t2v_low_noise_14B_fp8_scaled.safetensors
• Load CLIP → Wan_2.2_ComfyUI_Repackaged/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors
• Load VAE → Wan_2.2_ComfyUI_Repackaged/split_files/vae/wan_2.1_vae.safetensors
• EmptyHunyuanLatentVideo → set width / height / length
• CLIP Text Encoder → fill in the positive / negative prompt
• Click Run

4. 14B Image-to-Video (I2V)

4.1 Download the Workflow

1

wget https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/video_wan2_2_14B_i2v.json

Then click Menu -> File -> Open to import the workflow.

Or click Menu -> Templates -> search for Wan2.2 14B I2V in turn to load the workflow.

4.2 Generate the Video

• Download the example image

1

wget https://raw.githubusercontent.com/Comfy-Org/example_workflows/refs/heads/main/video/wan/2.2/input.jpg

• The first Load Diffusion Model → Wan_2.2_ComfyUI_Repackaged/split_files/diffusion_models/wan2.2_i2v_high_noise_14B_fp16.safetensors
• The second Load Diffusion Model → Wan_2.2_ComfyUI_Repackaged/split_files/diffusion_models/wan2.2_i2v_low_noise_14B_fp16.safetensors
• Load CLIP → Wan_2.2_ComfyUI_Repackaged/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors
• Load VAE → Wan_2.2_ComfyUI_Repackaged/split_files/vae/wan_2.1_vae.safetensors
• Load Image → upload the start frame
• CLIP Text Encoder → describe the desired motion changes
• EmptyHunyuanLatentVideo → set the dimensions / number of frames (optional)
• Click Run

5. 14B First-Last-Frame Video (FLF2V)

First-last frame: given a start frame and an end frame, the model automatically generates the transition animation in between. The model is the same as 14B I2V, so no additional download is needed.

5.1 Download the Workflow

1

wget https://raw.githubusercontent.com/Comfy-Org/workflow_templates/refs/heads/main/templates/video_wan2_2_14B_flf2v.json

Then click Menu -> File -> Open to import the workflow.

Or click Menu -> Templates -> search for Wan2.2 14B FLF2V in turn to load the workflow.

5.2 Download the Example First/Last Frame Assets (Optional)

• Download the example first/last frame assets

1
2

wget https://raw.githubusercontent.com/Comfy-Org/example_workflows/refs/heads/main/video/wan/2.2/wan22_14B_flf2v_start_image.png
wget https://raw.githubusercontent.com/Comfy-Org/example_workflows/refs/heads/main/video/wan/2.2/wan22_14B_flf2v_end_image.png

• The first Load Image → upload the start frame
• The second Load Image → upload the end frame
• WanFirstLastFrameToVideo → set width / height (try 720P if you have enough VRAM)
• CLIP Text Encoder → describe the transition motion
• Click Run