Installing Stable Swarm UI on Windows for Quick SDXL Images

Requirements:

•  nVidia's CUDA Toolkit
• Python, which can be found in the Windows Store or on the Python web site
• Install PyTorch by running correct installation command for your system, which can be found here

Steps:

• Check requirements are installed         
  python --version         
nvcc --version         
python -c "import torch; print(torch.__version__)"
• Download the repo: https://github.com/Stability-AI/StableSwarmUI         
  This can be done by clicking the green "Code" button in the upper right corner and selecting "Download ZIP" to get the latest version, or check the Releases page for a more stable version.
• Extract the zip file.
• Run the file install-windows.bat
• When the installation is complete, run the file launch-windows.bat to start the initial configuration setup.
• In the first three steps: agree to the EULA, select your preferred theme and choose who gets to use the installed app.
• In the fourth step, select ComfyUI to run everything locally.
• In step 5, select the stable diffusion models. We want to generate SDXL images, so we selected the SDXL base and SDXL refiner model.
• In the last step, click Install Now to begin the installation. You can monitor the installation progress in the Web UI.
• Before we can start generating images, we need to make sure ComfyUI has all the requirements to run.
  • Navigate to the folder "dbackend/comfy/ComfyUI"
  • Type Powershell in the address bar to open Powershell in that folder. (Or just cd to the folder in Powershell).
  • Run the command pip install -r requirements.txt
  • Optional: Run the command python main.py    
    The URL on which ComfyUI is running should be shown in the output. Visit that link to check ComfyUI is working. Then, if you can resist playing with Comfy, close the browser tab and hit Ctrl+C in Powershell to stop the server.
• Back in the main instalation folder, run  launch-windows.bat to start the app.

Generating an image:

• Select a model from the Model dropdown menu. We'll select the SDXL model. If you want weird and very crisp results, it's possible to select the SDXL Refiner as the base model here.
• Fill in the parameters in the form on the left hand side. 
• Click Generate. Wait for the result.