Overview
This video provides a beginner-friendly, step-by-step guide to installing Stable Diffusion on a new Windows computer, including prerequisites, model selection, troubleshooting tips, and advice on updates.
Prerequisites and System Requirements
- Installation is designed for Windows users with an Nvidia GPU (4GB+ VRAM recommended; 2GB minimum).
- Not suitable for AMD GPU or Apple/Linux users, but alternative guides are available via provided links.
- Essential prerequisites: Python 3.10.6 and Git for Windows.
Installing Python and Git
- Download Python 3.10.6 (64-bit Windows installer) and ensure "Add Python to PATH" is selected before installing.
- Download Git (64-bit Windows setup), choose preferred editor (e.g., Notepad++), and complete installation.
Preparing Stable Diffusion Folder
- Create a new folder for Stable Diffusion (name without spaces recommended).
- Open Command Prompt in this folder and run the provided command to clone the web UI repository.
Downloading and Organizing Models
- Official models include versions 1.4, 1.5, inpainting model, 2.0, and 2.1 (with various sub-versions).
- Recommended models for beginners: 1.4, 1.5, and inpainting model; optional: community models like Protogen V2.2.
- Download model files and, if using 2.1, also download the necessary .yaml configuration file.
- Move downloaded models and .yaml file into the appropriate "models" folder.
- Rename the .yaml file for 2.1 as needed to match the model type.
Configuring and Launching Stable Diffusion
- Edit the web UI’s user.bat file to set the Python path (copy Python executable path and paste after
set python=).
- Launch the web UI by double-clicking the user.bat file; installation may take 15–40 minutes depending on hardware.
- Access Stable Diffusion via the local URL provided once setup is complete.
Using Stable Diffusion Web UI
- Select the desired checkpoint/model, enter a prompt, set parameters, and click "Generate" to create AI art.
- Switch between models easily within the interface and adjust image dimensions for higher-res models.
- For 2.0 or 2.1 models, use negative prompts as needed for better results.
Troubleshooting and Optimization
- If issues occur with 2.1, edit the user.bat file to add
--xformers and restart to enable faster image generation.
- Comprehensive FAQ and community support are available through links and Discord.
Updating Stable Diffusion
- Manual update: open Command Prompt in the folder, run
git pull to fetch latest updates.
- Automatic update: edit user.bat file to add
git pull before launching, so updates run every time.
Recommendations / Advice
- Best models for beginners: 1.4, 1.5, and inpainting model for ease of use and available resources.
- For better images, try community models like Protogen V2.2.
- Skip 2.0 and use 2.1 if you want later model features.
- Be patient during installation and consult FAQ or Discord for help if needed.