Main

h1. ComfyUI Setup

The following describes how to setup "https://www.comfy.org/":ComfyUI on Ubuntu 24.04.  Note that a modern GPU is required.

h2. Install ComfyUI

ComfyUI will be installed into a Python virtual environment which keeps it somewhat isolated from the rest of the system.

# Install virtualenv if it is not already installed:

<pre>

$ sudo apt  install python3.12-venv

</pre>

# Create virtualenv at @/opt/comfyui@ and then update pip:

<pre>

$ sudo python3 -m venv /opt/comfyui/

$ sudo /opt/comfyui/bin/pip install --upgrade pip

</pre>

# Install pytorch into the virtualenv (NOTE: this assumes an NVidia GPU and the appropriate drivers must already be installed):

<pre>

$ sudo /opt/comfyui/bin/pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu121

</pre>

# Install ComfyUI within the virtualenv:

<pre>

$ cd /opt/comfyui

$ sudo mkdir comfyui

$ sudo chown -R dsorber:dsorber comfyui/

$ cd comfyui/

$ git clone https://github.com/comfyanonymous/ComfyUI.git .

$ sudo /opt/comfyui/bin/pip install -r requirements.txt 

</pre>

# Download and install Flux model

** https://huggingface.co/Comfy-Org/flux1-schnell/blob/main/flux1-schnell-fp8.safetensors

# Copy model to @models/checkpoints@ directory:

<pre>

mv ~/Downloads/flux1-dev-fp8.safetensors /opt/comfyui/comfyui/models/checkpoints

</pre>

# Base ComfyUI and Flux model is now installed. To run:

<pre>

$ cd /opt/comfyui/comfyui

$ ../bin/python3 main.py 

</pre>

h2. Resources

* https://comfyui-wiki.com/install/install-comfyui/install-comfyui-on-linux

* https://comfyui-wiki.com/tutorial/advanced/flux1-comfyui-guide-workflow-and-examples