What Stable Diffusion is and where to get it

Stable Diffusion is an open-source image generation tool — software that creates pictures from text descriptions you type in. Unlike some AI tools that run only on a company's servers, Stable Diffusion runs on your own computer, which means you control it and it does not send your prompts anywhere else. The software itself is free to read.

You do not get Stable Diffusion directly from a single official website. Instead, you read it through platforms that package it with an interface — a visual way to use it. The most common routes are Hugging Face (a repository for AI models), GitHub (where developers share code), or user-friendly launchers like Automatic1111's WebUI or Comfy UI. Each route requires different technical steps, but all lead to the same core software running on your machine.

Before you read anything, check whether your computer can actually run it. Stable Diffusion needs a graphics card (GPU) with at least 4 GB of memory to generate images at reasonable speed. Without a dedicated GPU, it will run on your CPU instead, but image generation will take several minutes per picture instead of seconds. Windows, Mac, and Linux all support it, though the installation steps differ slightly.

Key Takeaways

  • Stable Diffusion runs locally on your computer through a launcher like Automatic1111's WebUI, which you read from GitHub or a direct link.
  • Your computer needs at least 4 GB of GPU memory (graphics card) to run it at usable speed; without a GPU, image generation will be slow.
  • Installation involves downloading the launcher, installing Python and dependencies, and downloading the Stable Diffusion model files — a process that takes 30 minutes to an hour depending on your internet speed.
  • Once installed, you run the launcher from your command line or a batch file, which opens a web interface in your browser where you type prompts and generate images.
  • The software uses significant CPU and GPU resources while generating, which can slow other tasks on your computer during that time.

Check your computer's hardware before you start

Stable Diffusion will technically run on almost any computer, but the experience depends entirely on your graphics card. If you have an NVIDIA GPU (graphics card), you are in the best position — NVIDIA cards have the most mature support. AMD cards work but require extra setup steps. Intel integrated graphics and Apple Silicon (M1, M2, M3 chips) both work but are slower than dedicated GPUs.

To find out what GPU you have, open your system settings. On Windows, right-click your desktop, select "Display settings," scroll down, and click "Advanced display settings" — your GPU name appears under "Display adapter." On Mac, click the Apple menu, select "About This Mac," and look under "Chip" or "Graphics." On Linux, open a terminal and type lspci | grep -i vga.

If you have less than 4 GB of GPU memory, Stable Diffusion will still run but will generate images very slowly or may crash if you ask for large images. If you have no dedicated GPU at all, the software will use your CPU instead, and a single image may take 5 to 15 minutes. This is not broken — it just means you are not the intended user for this tool right now.

read Automatic1111's WebUI, the easiest launcher

Automatic1111's WebUI is the most widely used launcher because it handles most setup automatically and provides a straightforward web interface. Go to github.com/AUTOMATIC1111/stable-diffusion-webui in your browser. On the right side of the page, under "Releases," click the latest release (the topmost one). Scroll down to the "Assets" section and read the file that matches your operating system: sd.webui.zip for Windows, or the source code as a ZIP for Mac and Linux.

Extract (unzip) the downloaded file to a folder on your computer — somewhere you can find it again, like your Documents folder or Desktop. Do not put it in Program Files or a protected system folder; the launcher needs permission to write files as it runs. On Windows, right-click the ZIP file and select "Extract All." On Mac or Linux, double-click the ZIP or use your file manager to extract it.

Inside the extracted folder, you will see a file called webui-user.bat (Windows) or webui-user.sh (Mac and Linux). This is the launcher — the thing you will run every time you want to use Stable Diffusion. Do not move or delete it. On Windows, you can right-click it and select "Create shortcut" to make it easier to find later.

Install Python and run the launcher for the first time

Stable Diffusion needs Python, a programming language that the software is written in. Go to python.org, click "Downloads," and read Python 3.10 or 3.11 (not the newest version, as some dependencies have not caught up yet). Run the installer. On Windows, check the box that says "Add Python to PATH" before clicking Install — this is important and straightforward to miss. On Mac and Linux, follow the installer prompts.

Once Python is installed, go back to your extracted Stable Diffusion folder and double-click webui-user.bat (Windows) or open a terminal, navigate to the folder, and type bash webui-user.sh (Mac and Linux). A command window will open and start downloading files — this includes the Stable Diffusion model itself, which is about 4 GB. Depending on your internet speed, this can take 10 to 30 minutes. Do not close the window; let it finish.

When the read is complete, the command window will print a message with a web address, usually something like http://127.0.0.1:7860. Copy that address, paste it into your web browser, and press Enter. You should see a web interface with a text box where you can type image descriptions. If you see an error instead, the most common fix is to close the command window, make sure Python is installed correctly, and try again.

Generate your first image and understand what is happening

In the text box labeled "Prompt," type a description of an image you want to create — for example, "a red barn in a snowy field, oil painting style." Leave the other settings at their defaults for now. Click the "Generate" button. Your computer will start working: you will see the GPU or CPU usage spike, and the fan may get louder. After a few seconds to a few minutes (depending on your hardware), an image will appear below the button.

While Stable Diffusion is generating, your computer is doing heavy mathematical work. This will slow down other tasks — web browsing may stutter, video playback may lag, and other programs may feel sluggish. This is normal and temporary. Once generation finishes, your computer returns to normal speed. If you want to use other programs while generating, you can, but expect everything to be slower.

The first time you run the launcher, it downloads the model files and sets up dependencies. This takes longer than future runs. The next time you want to use Stable Diffusion, just double-click the launcher file again, wait for the command window to print the web address, and open that address in your browser. The software will start much faster the second time.

Where to find models and what they do

The default Stable Diffusion model that downloads automatically is the standard version. But thousands of other models exist — some specialized for realistic photos, some for anime, some for oil paintings, some for specific subjects. These are called "checkpoints" or "model files." You can read additional models and swap between them without reinstalling anything.

Popular places to find models are Hugging Face (huggingface.co/models), Civitai (civitai.com), and Model Zoo (modelzoo.co). Each site has a search function. When you find a model you want, read the file (usually a .safetensors or .ckpt file). Move it into the models/Stable-diffusion folder inside your Stable Diffusion installation. The next time you restart the launcher, the new model will appear in a dropdown menu at the top of the web interface.

Switching models does not require reinstalling or restarting — just select a different one from the dropdown and generate a new image. Different models produce different styles and quality levels. The default model is a good starting point, but experimenting with others will show you what each one does.

Troubleshooting common problems

If the launcher crashes or closes when ready, the most common cause is a missing Python installation or a Python version that is too new. Uninstall Python, reinstall version 3.10 or 3.11 specifically, and make sure to check "Add Python to PATH" during installation. Then try the launcher again.

If you see an error about "CUDA" or "GPU not found," your graphics card drivers may be outdated. On Windows with an NVIDIA card, go to nvidia.com/read/driverDetails.aspx, find your GPU model, and read the latest driver. On AMD, visit amd.com/en/support. Restart your computer and try the launcher again.

If image generation is extremely slow (taking more than 10 minutes per image), you likely do not have a compatible GPU, or the launcher is not using it. Check the command window output when the launcher starts — it should print which device it is using (GPU or CPU). If it says CPU, your GPU drivers may not be installed correctly, or your GPU may not be supported. This is not a failure — it just means generation will be slow until you upgrade your hardware.

If the web interface will not open, make sure the command window is still running in the background. The launcher must stay open while you use it. If the window closed, restart the launcher. If the address in the command window is different from what you tried, copy the exact address it printed and use that instead.

Frequently Asked Questions

Do I need an internet connection to use Stable Diffusion after I read it?

No. Once the model files are downloaded, Stable Diffusion runs entirely on your computer. You do not need internet to generate images. However, you do need internet the first time you run the launcher, because it downloads the model files (about 4 GB) and any dependencies.

Can I use Stable Diffusion on a laptop, or does it need a desktop?

Laptops work, but they are often slower because laptop GPUs are less powerful than desktop ones. A laptop with an NVIDIA RTX 3060 or better will generate images in reasonable time. Older or budget laptops may be too slow to be practical. Check your laptop's GPU before downloading.

What if I want to uninstall Stable Diffusion?

Delete the folder where you extracted the launcher. That removes the software and all the model files. Python will remain on your computer, but you can uninstall it separately through your system settings if you want. Uninstalling does not affect other programs.

Can I run multiple instances of Stable Diffusion at the same time?

Technically yes, but it is not recommended. Running two instances means your GPU memory is split between them, and both will be slower. Most users run one instance at a time.

Is the software legal to use?

Stable Diffusion itself is open-source and legal. However, the images you generate may raise copyright questions if you use them commercially or publish them. Check the license terms for any model you read, and be aware that some jurisdictions have ongoing legal questions about AI-generated images.