markdowndocs

View project on GitHub

up

Intro

github - (https://github.com/comfyanonymous/ComfyUI)

Installation

  • you will need the right python venv - currently 3.12 
    sudo add-apt-repository ppa:deadsnakes/ppa
sudo apt update
sudo apt install python3.12
sudo apt install python3.12-dev
sudo apt install python3.12-venv
python3.12 --version

    Windows

  • You can install multiple versions of python on windows by default
  • If you need an old version
    • Go get an AMD64 installer msi of the right version
    • an old version (the last one in the series that has an installer and not just a security update)
    • For example python 3.10.11 is here: (https://www.python.org/downloads/release/python-31011/) = For example python 3.12.10 is here: (https://www.python.org/downloads/release/python-31210/
    • Make sure you click “Add Python to path” at installation time
  • py --list will list what versions you have available
  • py -3.12 -m venv .venv to install it
  • .venv\Scripts\activate

Then do the following

  • Clone the repository:
    • git clone https://github.com/comfyanonymous/ComfyUI.git
  • Navigate to the directory: cd ComfyUI
  • Create a virtual environment:
    • python3.12 -m venv venv
  • Activate the environment:
    • source venv/bin/activate on Linux or MacOS
    • venv\Scripts\activate on Windows
  • Install dependencies:
    • Install PyTorch with CUDA or ROCm support according to your GPU. The exact commands vary, so consult the official ComfyUI installation instructions or relevant guides for your system.
    • Windows CUDA Archive - Old Versions: (https://developer.nvidia.com/cuda-toolkit-archive)
    • pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
    • pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu124
    • pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu128
    • As of June 2025, the official pre-built PyTorch wheels only support up to CUDA 12.8.
    • Test with python; import torch; torch.cuda.is_available()
  • Install other dependencies:
    • pip install -r requirements.txt
  • Download models:
    • Place the necessary Stable Diffusion models (e.g., checkpoints) in the ComfyUI/models/checkpoints directory.
  • Launch ComfyUI:
    • python3 main.py

Misc Repos

  • ComfyUI - git clone https://github.com/comfyanonymous/ComfyUI.git
  • ComfyUI Node Manager - git clone https://github.com/ltdrdata/ComfyUI-Manager.git
  • WD14 Reverse Tagger - git clone https://github.com/pythongosssss/ComfyUI-WD14-Tagger
  • image2prompt - git clone https://github.com/zhongpei/Comfyui_image2prompt

Installing Node Manager

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

Installing

Models

  • Reside on Hugging face or civitai
  • Best way to find them is to ask Perplexity as HF search often doesn’t workcd

CIVITAI

  • https://civitai.com/
  • (username redacted)

Tensor.Art

  • https://tensor.art/

uv Installation

  • got it out of a perplexity space
  • uv notes - https://codemaker2016.medium.com/introducing-uv-next-gen-python-package-manager-b78ad39c95d7

Step-by-step

  • Clone Comfy UI into your directory - git clone https://github.com/comfyanonymous/ComfyUI.git comfyui0
  • Change into rep directory = cd confyui0
  • Create venv - uv venv
  • Install the appropriate versoin of Pytorch - uv pip install torch torchvision torchaudio --torch-backend=cu128
    • .venv\Scripts\Activate
    • python -c "import torch; print(torch.cuda.is_available())"
    • If it is not True, you need to fix it.
  • Install ComfyUI dependencies - uv pip install -r requirements.txt
    • Maybe this happens automagicaly when we installed torch?
  • Optionally uv lock it - uv lock
    • This creates a uv.lock file with all the dependencies in it
  • Run it - uv run python main.py