How to Set Up Stable Diffusion on Ubuntu: A Complete Guide

Generating high-quality AI imagery locally offers unmatched privacy, zero subscription fees, and complete control over your creative workflow. For Linux power users, Ubuntu serves as the gold-standard operating system for running Stable Diffusion due to its robust driver support and efficiency with Python-based environments.

Hardware Requirements for Stable Diffusion on Ubuntu

Before diving into the installation, it is critical to ensure your workstation can handle the computational load of deep learning models. Stable Diffusion relies heavily on your GPU’s VRAM (Video RAM) to process tensors.

Note: While AMD and Intel GPUs can run Stable Diffusion, NVIDIA GPUs using CUDA remain the industry standard for compatibility and performance.

Preparing Your Ubuntu Environment

The installation process requires a clean environment to avoid dependency conflicts. We will use the Automatic1111 WebUI, which is the most widely supported interface for local Stable Diffusion setups.

1. Update and Upgrade

Open your terminal and ensure your package lists and installed packages are current: “bash sudo apt update && sudo apt upgrade -y “

2. Install Dependencies

You need specific system-level packages to compile and run the necessary libraries. Run the following command: “bash sudo apt install wget git python3 python3-venv libgl1 libglib2.0-0 -y “

3. Install NVIDIA Drivers

If you have not already installed proprietary drivers, use the ubuntu-drivers utility to find the best version for your hardware: “bash ubuntu-drivers devices sudo apt install nvidia-driver-535 # Replace with the recommended version “ Reboot your system after the driver installation to ensure the kernel recognizes the GPU.

Step-by-Step: How to Set Up Stable Diffusion on Ubuntu

With the system prepped, we can now clone the repository and initialize the environment.

Cloning the Repository

Navigate to your desired installation directory (e.g., ~/AI) and clone the Automatic1111 repository: “bash mkdir ~/AI cd ~/AI git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui cd stable-diffusion-webui “

Creating the Virtual Environment

Using a Python virtual environment keeps your global system files clean: “bash python3 -m venv venv source venv/bin/activate “

Launching the Installation

When you run the launch script for the first time, the installer will automatically download the base Stable Diffusion models and install all required Python packages (PyTorch, Torchvision, etc.).

“bash ./webui.sh “ Be patient. This process may take several minutes depending on your internet connection speed, as it downloads several gigabytes of model weights.

Optimizing Stable Diffusion Performance

Once the installation finishes, the script will provide a local URL (usually http://127.0.0.1:7860). However, you can optimize the startup for your specific hardware by editing the webui-user.sh file.

Open the file in a text editor: “bash nano webui-user.sh “

Modify the COMMANDLINE_ARGS line. Common optimizations include:

For 4GB-6GB VRAM: Use --medvram to offload tasks to system RAM.
For low VRAM/Integrated GPUs: Use --lowvram.
To enable Xformers: Add --xformers to significantly speed up image generation.

Example: “bash export COMMANDLINE_ARGS="--xformers --medvram" “

Troubleshooting Common Setup Issues

Even with a perfect setup, you may encounter obstacles. Here is how to handle the most frequent ones:

CUDA Out of Memory

If you see a “CUDA Out of Memory” error, your GPU is attempting to generate an image size larger than its VRAM capacity. Reduce your resolution settings (e.g., from 1024×1024 to 512×512) or ensure you have added the --medvram flag to your launch arguments.

Python Version Mismatch

Stable Diffusion requires specific versions of Python. If the script fails to start, verify your version by running python3 --version. If it is older than 3.10, you may need to install a newer version via the deadsnakes PPA.

Permission Denied

If you encounter permission errors, ensure your user owns the folder: “bash sudo chown -R $USER:$USER ~/AI/stable-diffusion-webui “

FAQ

Is it free to use Stable Diffusion on Ubuntu?

Yes. Stable Diffusion is open-source software. Aside from the electricity costs and your hardware investment, there are no subscription fees or per-image generation costs.

Can I run Stable Diffusion on a CPU?

While possible, it is not recommended. Running on a CPU is significantly slower than on a GPU. If you lack an NVIDIA card, you may need to use specific launch arguments like --use-cpu all, but expect generation times to increase from seconds to minutes per image.

How do I update the WebUI?

Because the software is managed via Git, updating is simple. Just navigate to the stable-diffusion-webui folder and run git pull. The next time you run webui.sh, it will pull the latest features and fixes.

Does this setup work on Ubuntu Server?

Yes, but you will need to access the WebUI from a different machine on your local network. Use the --listen command line argument to allow external connections to your Ubuntu server’s IP address.

Conclusion

Setting up Stable Diffusion on Ubuntu provides a robust, professional-grade platform for AI art generation. By following this guide, you have moved beyond simple cloud-based interfaces and gained full control over your models, extensions, and hardware performance. As you become more comfortable with the environment, explore community-made checkpoints on platforms like Civitai to further customize your artistic output.