Hello, I wanted to put together an article series on Stable Diffusion techniques.

What is SD?

Stable Diffusion is software for generating various kinds of art. Common operating modes include:

  • txt2img - Put in a prompt (e.g. “woman standing in a field”), get an image back
  • img2img - Put in an image and a prompt, get a new image back
  • txt2vid - Put in a prompt, get a video back
  • vid2vid - Put in a video and a prompt, get a new video back

Each mode has different required inputs and configuration options which can be used to tweak/customize the output.

Installing webui

  1. Download this repo to your computer using Git or the “Download ZIP” button on GitHub (extract the ZIP after downloading it).
  2. Download the Stable Diffusion 1.5 model. Place the safetensors file in <webui>/models/stable-diffusion/.
  3. In the webui folder, look for webui-start.sh (Linux/Mac) or webui-start.bat (Windows). Double-click it to start webui. On first startup, it will download and install all required dependencies.
  4. When the terminal window says “Running on local URL: http://127.0.0.1:7860” you can open a browser window and navigate to http://127.0.0.1:7860 - you should see the SD webui in your browser.
  5. Congrats, you’re done!

Upgrading webui

If you used Git to get webui, then run the following in the webui directory:

git pull

Now restart webui. New dependencies will be installed if required.

If you used a ZIP file to get webui, simply extract the new ZIP on top of the existing webui directory.

Using webui

Now you’ve got the latest and greatest Stable Diffusion UI up and running. The next article in the series will cover the common operating modes, then we’ll go over popular extensions, and finish things off with some more advanced techniques.