The NOOB ComfyUI Install Guide

Way easier than other UIs.

Windows + Nvidia

If you are on windows with nvidia it's literally just download + extract:
https://github.com/comfyanonymous/ComfyUI#installing

Then download this (it's a SD checkpoint):
https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.ckpt

Open up the dir you just extracted and put that v1-5-pruned-emaonly.ckpt file in ComfyUI\models\checkpoints

Then run ComfyUI using the bat file in the directory.

Then press "Queue Prompt" in the UI and wait a bit.

And you have your first image.

Linux + AMD

git clone https://github.com/comfyanonymous/ComfyUI
cd ComfyUI
python -m pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/rocm5.6 -r requirements.txt

#optional, download a checkpoint
wget -c https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.ckpt -P ./models/checkpoints/

python main.py

#If that fails because your GPU isn't officially supported by ROCm (6700 XT , etc..) you can try:
HSA_OVERRIDE_GFX_VERSION=10.3.0 python main.py

For Installing on Nvidia + Linux and others click here

After installing you can check out the examples by clicking here

ComfyUI supports loading the workflow from PNG files that were generated with ComfyUI so load the PNGs on the examples page and have fun.

Edit
Pub: 17 Mar 2023 01:53 UTC
Edit: 06 Oct 2023 15:02 UTC
Views: 73058