How to Install and Use Stable Diffusion on Your Breeze

How to Install and Use Stable Diffusion on Your Breeze

Stable Diffusion is an open-source text-to-image AI model that generates high-quality images from text prompts. Running it on your own Breeze gives you full control over generation, no per-image costs, and complete privacy for your creative work.

Prerequisites

• A Breeze instance with a GPU (recommended: at least 8 GB VRAM) or a high-RAM CPU instance for slower generation
• Ubuntu 22.04 or later
• At least 20 GB of free disk space for model weights
• Python 3.10 or later

Installing System Dependencies

Update your system and install required packages:

sudo apt update && sudo apt upgrade -y
sudo apt install -y python3 python3-pip python3-venv git wget libgl1 libglib2.0-0

If your Breeze has an NVIDIA GPU, install the appropriate CUDA drivers following the NVIDIA documentation for your GPU model.

Setting Up the Web UI

The most popular interface for Stable Diffusion is the Automatic1111 Web UI. Clone and set it up:

cd ~
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
cd stable-diffusion-webui

For CPU-only Breeze instances, edit webui-user.sh and add:

export COMMANDLINE_ARGS="--skip-torch-cuda-test --no-half --use-cpu all --listen"

For GPU instances, simply add --listen to make it accessible remotely.

Downloading Model Weights

Download a Stable Diffusion model checkpoint and place it in the models directory:

cd ~/stable-diffusion-webui/models/Stable-diffusion
wget -O sd_v1-5.safetensors "https://huggingface.co/runwayml/stable-diffusion-v1-5/resolve/main/v1-5-pruned-emaonly.safetensors"

You can also download SDXL or SD 3.x models for higher quality output, though these require more VRAM.

Launching the Web UI

Start the application:

cd ~/stable-diffusion-webui
./webui.sh

The first launch will download and install all Python dependencies automatically. Once ready, access the interface at http://your-breeze-ip:7860.

Generating Your First Image

In the web interface, enter a prompt such as “a futuristic cityscape at sunset, digital art, highly detailed” and click Generate. Adjust these key parameters:

• Steps — higher values (20–50) produce more detailed images but take longer
• CFG Scale — controls how closely the image follows your prompt (7–12 is typical)
• Sampler — Euler a or DPM++ 2M Karras are good defaults
• Resolution — 512x512 for SD 1.5, 1024x1024 for SDXL

Running as a Background Service

Create a systemd service so Stable Diffusion starts automatically and survives reboots:

[Unit]
Description=Stable Diffusion Web UI
After=network.target

[Service]
User=deploy
WorkingDirectory=/home/deploy/stable-diffusion-webui
ExecStart=/bin/bash webui.sh
Restart=always

[Install]
WantedBy=multi-user.target

Security Considerations

By default the web UI has no authentication. Place it behind an Nginx reverse proxy with basic auth or use the built-in --gradio-auth user:password flag to restrict access on your Breeze.