Stable Diffusion on macOS with Metal Acceleration

Run Stable Diffusion XL (SDXL) natively on Apple Silicon Macs (M1 / M2 / M3) using PyTorch with Metal (MPS) acceleration.

This project is based on 🤗 Hugging Face diffusers and uses a simple generate.py script to prompt and generate images locally on macOS.

---

🚀 What This Project Does

• Runs Stable Diffusion locally on macOS
• Uses Apple Metal (MPS) instead of CUDA
• Prompts for text input directly in the terminal
• Saves generated images automatically
• Works offline after first model download

---

🔧 Requirements

• macOS 26+
• Apple Silicon (M1, M2, or M3)
• Python 3.10+
• Xcode Command Line Tools
• pip and virtualenv

Install Xcode CLI tools if needed:

bash

xcode-select --install

⚙️ Setup Instructions

Clone this repository

git clone https://github.com/El1Leon/Stable_Diffusion_macOS_Metal.git cd Stable_Diffusion_macOS_Metal

Create and activate a virtual environment

python3 -m venv .venv source .venv/bin/activate

Upgrade pip

pip install --upgrade pip

Install dependencies

pip install -r requirements.txt

🖼️ Generating Images with generate.py

Once everything is installed, run:

bash

python generate.py