What you're downloading and where it lives

Stable Diffusion is an AI image generator that runs on your own computer instead of through a website. When you download it, you're getting software that turns text descriptions into images — but the actual program lives locally on your machine, not on someone else's server. This means you control it completely, but you also need enough disk space and processing power to run it.

The most common way to get Stable Diffusion is through a launcher called Automatic1111, which is a free interface that handles all the technical setup for you. Without it, you'd be typing commands into a terminal window. Automatic1111 gives you buttons and text boxes instead. There are other launchers available — Comfy UI and InvokeAI are popular alternatives — but Automatic1111 is the most straightforward starting point for someone new to this.

Before you start downloading anything, check whether your graphics card can handle it. Stable Diffusion runs much faster on an NVIDIA GPU than on a CPU alone. If you have an NVIDIA card, the process is faster. If you have an AMD card or only integrated graphics, the software still works, but image generation will take longer — sometimes 5 to 10 minutes per image instead of 30 seconds.

Key Takeaways

  • Stable Diffusion runs on your computer through a launcher like Automatic1111, which you download from GitHub as a folder you extract and run locally.
  • You need Python installed first, plus about 10 to 25 GB of free disk space depending on which model files you choose to download.
  • NVIDIA graphics cards speed up image generation significantly; AMD and Intel integrated graphics work but are much slower.
  • The first time you run Automatic1111, it downloads the core model files automatically, which takes 10 to 30 minutes depending on your internet speed.
  • Once running, you access Stable Diffusion through a web browser window on your own machine — no account or internet connection required after setup.

Installing Python and checking your system

Stable Diffusion needs Python to run. Python is a programming language that acts as the foundation for the software. Go to python.org, click the Downloads button, and download the latest version (currently 3.11 or 3.12). Run the installer. When you see the option to "Add Python to PATH," check that box — this lets your computer find Python from anywhere.

After Python installs, open Command Prompt (Windows) or Terminal (Mac/Linux) and type python --version. If you see a version number back, Python is installed correctly. If you see "command not found," the PATH didn't set properly — uninstall Python and run the installer again, making sure to check that PATH box.

Next, check your graphics card. On Windows, right-click your desktop and look for NVIDIA Control Panel or AMD Radeon Settings. If you see either, you have a dedicated GPU. If you see neither, you're using integrated graphics built into your CPU, which will work but slowly. You don't need to do anything differently — just know that image generation will take longer.

Downloading Automatic1111 from GitHub

Go to github.com/AUTOMATIC1111/stable-diffusion-webui. On the right side of the page, you'll see a green button labeled "Code." Click it, then click "Download ZIP." This downloads the entire Automatic1111 folder as a single compressed file.

Once the ZIP file finishes downloading, right-click it and choose "Extract All" (Windows) or double-click it (Mac). Choose where you want the folder to live — your Desktop or Documents folder works fine. Do not put it inside Program Files or another protected folder, because the software needs permission to create new files inside it as it runs.

After extraction, you'll have a folder called stable-diffusion-webui-master. Open it. Inside, you'll see several files and folders. The ones you care about are the batch files on Windows (webui-user.bat) or the shell script on Mac/Linux (webui-user.sh). These are the files you'll run to start Stable Diffusion.

Running Automatic1111 for the first time

On Windows, double-click webui-user.bat. A Command Prompt window will open and start downloading files. On Mac or Linux, right-click webui-user.sh, choose "Open With" and select Terminal, or open Terminal yourself and drag the file into it.

The first run takes the longest. The software downloads Python dependencies (libraries it needs to function), then downloads the base Stable Diffusion model file, which is about 4 GB. Depending on your internet speed, this can take 10 to 30 minutes. You'll see text scrolling in the Command Prompt window. This is normal — do not close the window.

When the downloads finish, you'll see a message that says something like "Running on local URL: http://127.0.0.1:7860". Open your web browser and go to that address. You should see the Stable Diffusion interface — a text box where you type descriptions, a Generate button, and settings on the right side. If you see this, the installation worked.

Understanding disk space and model files

The base Stable Diffusion model is about 4 GB. If you want to use different models — different versions that produce different styles or quality levels — each one takes another 4 to 7 GB. You don't need to download them all at once. Start with the default model and add others later if you want variety.

Beyond the models themselves, the software and its dependencies take up another 5 to 10 GB. So a basic working setup needs about 10 GB free, and if you add a few extra models, you're looking at 20 to 30 GB total. Check your available disk space before you start. On Windows, right-click your C: drive and choose Properties. On Mac, click the Apple menu, choose About This Mac, and click Storage.

If you're running low on space, you can move the entire stable-diffusion-webui-master folder to an external drive or a different internal drive, and run it from there. Just make sure the drive is fast enough — a USB 2.0 drive will be slow, but a USB 3.0 drive or external SSD works fine.

Troubleshooting the most common problems

If the Command Prompt window closes immediately without downloading anything, Python probably isn't installed or isn't in your PATH. Go back to the Python installation step and run the installer again, checking the PATH box carefully. Then try running webui-user.bat again.

If you see an error about CUDA or GPU drivers, your NVIDIA drivers are out of date. Go to nvidia.com/Download/driverDetails.aspx, enter your graphics card model, and download the latest driver. Install it and restart your computer, then try running webui-user.bat again.

If the browser window won't open or shows "connection refused," the software is still downloading. Wait another 5 to 10 minutes and try refreshing the page. If it still doesn't work, check the Command Prompt window — if you see error messages in red text, take a screenshot and search for that exact error message online, because it usually points to a specific missing dependency.

If image generation is extremely slow (taking 10+ minutes per image on an NVIDIA card), check your graphics card memory. Some older cards have only 2 to 4 GB of VRAM. Stable Diffusion can still run on these, but you need to enable "Optimize Memory" in the Settings tab. This makes generation slower but prevents out-of-memory crashes.

What to do after your first image

Once you've generated your first image, you can explore the settings. The "Sampling method" dropdown changes how the AI generates images — different methods produce different results. The "Steps" slider controls how many refinement passes the AI makes; more steps usually mean better quality but longer generation time. Start with 20 to 30 steps and adjust from there.

If you want to try different models, go to the folder where you extracted Automatic1111, then navigate to models > Stable-diffusion. This is where model files live. You can download new models from huggingface.co or civitai.com — just download the .safetensors file and drop it in that folder. The next time you run Automatic1111, the new model will appear in the Model dropdown at the top of the interface.

The software stays running in your Command Prompt window as long as you're using it. When you're done, close the browser tab, then close the Command Prompt window. The next time you want to use Stable Diffusion, just double-click webui-user.bat again — it won't need to download the models a second time.

Frequently Asked Questions

Do I need an internet connection to use Stable Diffusion after it's installed?

No. After the initial setup downloads the model files, you can disconnect from the internet and Stable Diffusion will still work. The software runs entirely on your computer. You only need internet during the first installation and if you want to download additional models later.

Can I use Stable Diffusion on a Mac with an Apple Silicon chip?

Yes, but the setup is slightly different. Apple Silicon Macs use Metal acceleration instead of CUDA. When you run webui-user.sh, the software detects your Mac and configures itself automatically. Generation will be slower than on an NVIDIA GPU but faster than on older Intel Macs.

What if I want to uninstall Stable Diffusion?

Delete the stable-diffusion-webui-master folder. That's it. The software doesn't install anything system-wide, so there's nothing else to uninstall. If you want to free up space, you can also delete the models folder inside it, though you'll need to re-download models if you install Stable Diffusion again later.

Can I run Stable Diffusion on my laptop?

Yes, as long as it has at least 8 GB of RAM and 20 GB of free disk space. Laptops with integrated graphics will be slow — expect 5 to 15 minutes per image — but it works. Laptops with dedicated NVIDIA GPUs (like gaming laptops) will be much faster. Avoid running it on battery power; plug in your laptop because image generation uses a lot of power.

Is there a way to make Stable Diffusion faster?

If you have an NVIDIA card, make sure your drivers are up to date — outdated drivers significantly slow down generation. You can also reduce the image resolution (512x512 instead of 768x768) to speed things up. Lowering the Steps slider also makes generation faster but with lower quality. If you want the best speed, upgrading to a graphics card with more VRAM (8 GB or higher) makes the biggest difference.