A pretty front-end for Stable Diffusion

Installation

• Put your SD checkpoints (the huge ckpt / safetensors files) in models/checkpoints
• Put your VAE in models/vae

Installing ComfyUI on Windows

Portable builds for Window on NVidia GPUs (or CPU-only)

run_nvidia_gpu.bat

Installing ComfyUI on Linux

Installing to a Python venv

git clone git@github.com:comfyanonymous/ComfyUI.git
pyenv virtualenv 3.11.7 comfy
pyenv local comfy
pip install torch torchvision torchaudio \
  --extra-index-url https://download.pytorch.org/whl/cu121
pip install -r requirements.txt

The run it directly from the checkout

python main.py

ComfyUI-Manager

An extension designed to enhance the usability of ComfyUI

GitHub - ltdrdata/ComfyUI-Manager

Windows installation:

.\python_embeded\python.exe -s -m pip install gitpython
.\python_embeded\python.exe -c "import git; git.Repo.clone_from('https://github.com/ltdrdata/ComfyUI-Manager', './ComfyUI/custom_nodes/ComfyUI-Manager')"

Linux installation (from your existing ComfyUI checkout)

cd custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Manager.git

After installation, restart ComfyUI and select “Manager.” This is where you install models.

Related

• GitHub - comfyanonymous/ComfyUI: The most powerful and modular stable diffusion GUI with a graph/nodes interface.

Added to vault 2024-01-19. Updated on 2024-01-26