A simple script to make installing ComfyUI on Linux easier. This fork is intended solely as an installation script. It will create a new script (start.sh) in the chosen directory to update and launch ComfyUI once you've installed it.
Note that this script uses relative paths. I had no reason to change that so I left it as is, but just a heads up if you're running the script from another working directory.
Step 1) download the comfy-webui.sh file to your home directory.
Step 2) Open a terminal window and execute the next 2 commands.
chmod +x comfy-webui.sh <<-- set the executable permission
./comfy-webui.sh <<-- execute the script
Step 3) Lastly, after the script finishes, Press Ctrl-C to quit.
Step 4) Download models and place the files in the ComfyUI/models/checkpoints or ComfyUI/models/unet directories and VAE files in the ComfyUI/models/vae directory as normal. These can be found in a variety of places like civitai.com, huggingface.co, etc.
Step 5) Any time you wish to start the server, in your home directory, simply start ./start.sh.
ComfyUI Github Repository: https://github.com/comfyanonymous/ComfyUI
By using the start.sh script this creates to start the ComfyUI AI, your copy of ComfyUI will be automatically updated to the latest version of ComfyUI.
I wanted a script that would mostly automate the creation of a new ComfyUI instance in order to combat dependency conflicts by having instances depending on purpose (Image generation, inpainting, video generation, 3D mesh generation, background removal, etc.) and test new nodes/workflows/updates without breaking an existing installation.