Sign In
homemodelsimagesvideosarticlescomicschallengesupdatesshop

Sample Workflow

Kokemomo's Avatar

Kokemomo

Loading Images

⚠️Install VideoHelperSuite Custom Node to load workflow from videos.

🍎RTX4070 12G VRAM + 64G RAM, takes 4min ~ 1min for 480x640 77frames.

Installation Note

1.ComfyUI

🔗GitHub - comfyanonymous/ComfyUI

Download and extract ComfyUI Windows Portable Package.

Go to ComfyUI_windows_portable\ComfyUI\custom_nodes.

Open command line.

git clone https://github.com/kijai/ComfyUI-WanVideoWrapper.git
git clone https://github.com/Comfy-Org/ComfyUI-Manager.git
..\..\python_embeded\python.exe -m pip install -r ComfyUI-WanVideoWrapper\requirements.txt
..\..\python_embeded\python.exe -m pip install -r ComfyUI-Manager\requirements.txt

Go to ComfyUI_windows_portable.

Run run_nvidia_gpu.bat.

Download workflow image from above.

Drop workflow image file down to ComfyUI workspace.

ComfyUI Manager -> Install Missing Custom Nodes

Install them all (latest version).

-> Restart button

Wait requirements are installed.

2.Triton & SageAttention

Install CUDA12.9.

🔗CUDA Toolkit 12.9 Downloads | NVIDIA Developer

(CUDA12.8 also worked).

Install SageAttention2++.

🔗Release v2.2.0-windows · woct0rdho/SageAttention · GitHub

Download sageattention-2.2.0+cu128torch2.8.0-cp313-cp313-win_amd64.whl.

(depends on versions of libs that ComfyUI using. cp313 = python3.13, torch2.8 and CUDA 12.8)

Go to ComfyUI_windows_portable.

Put whl file on there.

python_embeded\python.exe -m pip install "sageattention-2.2.0+cu128torch2.8.0-cp313-cp313-win_amd64.whl"

Install Triton3.4.0.

Go to ComfyUI_windows_portable.

python_embeded\python.exe -m pip install triton-windows

🔗GitHub - woct0rdho/triton-windows: Fork of the Triton language and compiler for Windows support and easy installation

Download python_3.13.2_include_libs.zip.

Extract and copy "include" and "libs" to ComfyUI_windows_portable\python_embeded\.

Edit run_nvidia_gpu.bat.

.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build

to

.\python_embeded\python.exe -s ComfyUI\main.py --windows-standalone-build --use-sage-attention

Run run_nvidia_gpu.bat.

3.Models

🔗Kijai/WanVideo_comfy at main

  • to vae: Wan2_1_VAE_bf16.safetensors

  • to text_encoders: umt5-xxl-enc-bf16.safetensors

  • to diffusion_models: Wan2_1-I2V-14B-480P_fp8_e4m3fn.safetensors

🔗Comfy-Org/Wan_2.1_ComfyUI_repackaged at main

  • to clip_vision: clip_vision_h.safetensors

🔗ac-pill/upscale_models at main

  • to upscale_models: RealESRGAN_x4plus_anime_6B.pth

🔗Kijai/WanVideo_comfy at main

  • to loras: lightx2v_I2V_14B_480p_cfg_step_distill_rank64_bf16.safetensors

Comments