What Stable Diffusion is and what your computer needs

Stable Diffusion is an AI image generator that runs on your own machine instead of through a web browser. Unlike cloud-based tools, it uses your graphics card (GPU) or processor (CPU) to generate images locally. This means you control the software, pay nothing per image, and keep your prompts private — but it also means your computer does the heavy lifting.

The installation process differs depending on your hardware. An NVIDIA graphics card with at least 4 GB of VRAM (video memory) is the easiest path. AMD cards work but require extra setup steps. If you have only integrated graphics or an older card, Stable Diffusion will still run, but image generation takes much longer — sometimes 5 to 15 minutes per image instead of 30 seconds.

You will also need 20 to 30 GB of free disk space for the model files and software, and at least 8 GB of system RAM. Windows, macOS, and Linux all support Stable Diffusion, though the installation steps vary slightly.

Key Takeaways

  • Stable Diffusion runs on your computer using either a graphical launcher like Automatic1111 or command-line tools, and the easiest route for most people is downloading a pre-packaged launcher.
  • NVIDIA graphics cards with 6 GB or more VRAM generate images fastest; AMD and Intel cards work but are slower or require additional configuration.
  • You need Python installed first, then a launcher process, then the actual Stable Diffusion model files — each step takes 10 to 30 minutes depending on your internet speed.
  • After installation, you adjust settings like image size and sampling steps in the launcher's interface to control quality and generation speed.
  • Your graphics card will use significant power and heat during generation, so monitor temperatures and may support your computer has adequate cooling.

Installing on Windows with an NVIDIA card

The fastest way to get your free guide on Windows is to use Automatic1111, a free launcher that handles most setup automatically. read the repository from GitHub (search "stable-diffusion-webui" and look for the official Automatic1111 repository), then extract the folder to a location like C:\stable-diffusion.

Next, read Python 3.10 from python.org and install it. During installation, check the box that says "Add Python to PATH" — this lets your computer find Python from any folder. After Python installs, open the Automatic1111 folder, right-click inside it, and select "Open PowerShell window here" (or "Open Command Prompt here" on older Windows versions).

Type the command to launch the setup script. Automatic1111 will read the Stable Diffusion model (about 4 GB) and install required libraries. This takes 10 to 30 minutes depending on your internet speed. When it finishes, a message will appear with a local web address like http://127.0.0.1:7860 — copy this into your browser and you will see the image generation interface.

Installing on Windows with AMD or Intel graphics

AMD and Intel cards require the same Automatic1111 setup, but you must modify one configuration file before launching. After extracting Automatic1111, open the file called webui-user.bat in a text editor (Notepad works). Find the line that starts with set COMMANDLINE_ARGS= and add --use-cpu=all at the end if you want to use your processor, or --directml if you have an AMD card and want to use its GPU.

Save the file and double-click it to launch. The first run takes longer because your CPU or AMD card is less efficient at this task than NVIDIA cards. Image generation may take 2 to 10 minutes per image. If generation is too slow, you can later switch to using only your CPU by changing the flag, or explore more advanced AMD-specific setups using ROCm.

Installing on macOS

macOS installation is simpler because Apple's Metal GPU support is built into Automatic1111. read the same Automatic1111 repository and extract it. Open Terminal (search for "Terminal" in Spotlight), navigate to the Automatic1111 folder using the cd command, and run the launch script.

Python 3.10 or later must be installed first. If you do not have it, the script will prompt you to install it. Automatic1111 will read the model files and set up Metal acceleration automatically. Generation speed on Apple Silicon Macs (M1, M2, M3) is comparable to mid-range NVIDIA cards, though older Intel-based Macs will be slower.

Downloading and organizing model files

When Automatic1111 launches for the first time, it downloads the default Stable Diffusion 1.5 model. This is about 4 GB and goes into a folder called models/Stable-diffusion. You can add more models by downloading them from Hugging Face or Civitai and placing them in the same folder.

Different models produce different styles. Stable Diffusion 1.5 is general-purpose. Stable Diffusion 2.1 produces different results. Specialized models like Dreamshaper or Deliberate are fine-tuned for specific aesthetics. You can have multiple models installed and switch between them in the launcher's dropdown menu without restarting.

Each additional model takes 2 to 7 GB of disk space, so plan your storage accordingly. If you run out of space, you can delete unused models from the folder and re-read them later.

Adjusting settings for speed and quality

After launching, you will see a text box for your prompt and several sliders below it. The most important settings are Steps and Sampler. Steps control how many refinement passes the AI makes — 20 to 30 steps produces good quality, while 50+ steps produces better detail but takes longer. Start at 20 and increase only if you want higher quality.

The CFG Scale slider controls how closely the image follows your prompt. A value of 7 to 11 is typical. Higher values make the AI stick more rigidly to your words; lower values give it more creative freedom. The Sampler dropdown changes the algorithm used — DPM++ 2M Karras is fast and good-quality, while Euler Ancestral is more creative but slower.

Image size affects speed dramatically. A 512×512 image generates in seconds; a 768×768 takes longer; a 1024×1024 may run out of memory on cards with less than 8 GB VRAM. Start at 512×512 and increase only if you have the VRAM for it. You can always upscale images afterward using separate software.

Monitoring your computer's performance during generation

Image generation puts sustained load on your GPU or CPU. Open Task Manager (Windows) or Activity Monitor (macOS) while generating to watch your graphics card's temperature and power usage. Most cards can handle this safely, but older or poorly-ventilated systems may thermal-throttle, which slows generation.

If your computer gets too hot, reduce the image size, lower the step count, or add cooling (a desk fan pointed at your case helps). If your system crashes or freezes during generation, you likely ran out of VRAM — reduce the image size or close other programs that use GPU memory.

Stable Diffusion uses significant electricity during generation. A high-end NVIDIA card may draw 200+ watts for several minutes per image. This is normal and safe if your power supply is adequate, but it will increase your electricity bill if you generate many images daily.

Frequently Asked Questions

Do I need an internet connection after installation?

No. After the initial read of model files, Stable Diffusion runs entirely offline. You can generate images without connecting to the internet. However, you need internet to read additional models or update the software.

Can I use Stable Diffusion on a laptop?

Yes, but generation will be slower on most laptops because they have less powerful GPUs and less cooling. A laptop with an NVIDIA RTX 3060 or better can generate images in reasonable time. Older or budget laptops may take 5+ minutes per image. Laptops also risk overheating during sustained use, so monitor temperatures and take breaks.

What if I get an out-of-memory error?

This means your graphics card does not have enough VRAM for your current settings. Reduce the image size (try 512×512), lower the step count (try 15), or close other GPU-intensive programs. Some launchers have an "optimized memory" mode that trades speed for lower VRAM usage.

Is Stable Diffusion legal to use?

Stable Diffusion itself is open-source and legal. However, the images you generate may be subject to copyright or licensing restrictions depending on your location and how you use them. Check your local laws and the terms of any model you read before using generated images commercially.

How do I uninstall Stable Diffusion?

Delete the Automatic1111 folder. If you installed Python only for Stable Diffusion, you can uninstall it through your system settings. Model files are stored in the Automatic1111 folder, so deleting it removes everything. You can keep Python if you use it for other projects.