From Zero to Artist: ComfyUI Setup for M1, M2, and M3 Macs

This tutorial will guide you through installing ComfyUI on an Apple Silicon Mac (M1, M2, M3) with macOS 12.3 or later. The installation will utilize PyTorch’s Metal Performance Shaders (MPS) backend for GPU training acceleration.

Prerequisites

Before starting, ensure you have the following:

An Apple Silicon Mac (M1, M2, M3)
macOS 12.3 or later
Python 3.7 or later
Xcode command-line tools

Step-by-Step Installation

Step 1: Install Xcode Command-Line Tools

First, install the Xcode command-line tools:

xcode-select --install

Step 2: Install Miniconda

If you don’t already have Miniconda installed, follow the instructions in this guide on installing Conda on Apple Silicon Macs.

Step 3: Create and Activate a Conda Environment

Create a new Conda environment for ComfyUI:

conda create -n comfyui python=3.8
conda activate comfyui

Step 4: Install PyTorch Nightly with MPS Support

Install the latest nightly build of PyTorch with MPS support:

conda install pytorch torchvision torchaudio -c pytorch-nightly

Step 5: Clone the ComfyUI Repository

Clone the ComfyUI repository from GitHub into a directory of your choice:

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

Step 6: Install ComfyUI Dependencies

Install the required Python packages using pip within the Conda environment:

pip install -r requirements.txt

Step 7: Set Up Models

Create directories for your Stable Diffusion checkpoints and other models:

mkdir -p models/checkpoints models/vae models/embeddings

Downloading Checkpoint Files and VAE Models

1. Hugging Face Model Hub: The Hugging Face Model Hub provides a vast array of models, including Stable Diffusion. You can easily find and download checkpoints and VAE models from the Hugging Face Model Hub.

2. Civitai: Civitai is a popular community platform where users share various Stable Diffusion models and VAEs. You can explore and download the specific models you need from Civitai.

Copy your checkpoint files (e.g., .ckpt, .safetensors) and VAE models to the appropriate directories:

# Example commands
cp /path/to/downloaded/model.ckpt models/checkpoints/
cp /path/to/downloaded/model.safetensors models/checkpoints/
cp /path/to/downloaded/vae_model.ckpt models/vae/

Step 8: Run ComfyUI

Launch ComfyUI:

python main.py

Conclusion

By following these steps, you should have ComfyUI installed and running on your Apple Silicon Mac (M1, M2, M3) using the MPS backend for GPU acceleration. This setup ensures efficient performance and leverages the capabilities of your Mac’s hardware for advanced Stable Diffusion workflows.