The beauty of open source is that the passionate community behind this is always creating new stuff and sharing it with everyone. With an influx in AI generated art, there are many Google Colab Notebooks that are available for use to use. In this post I fill list all the favourites that I use and am aware of. So if are interested in running Stable Diffusion in your own Google Colab account (you can start with free one) just browse the list below and get the notebook that you want to play with
Deforum Stable Diffusion Notebook
Deforum Stable Diffusion is the most popular notebook available that allows you to run your own version of Stable Diffusion model. Make sure you download the Stable Diffusion weights from Hugging face website via the link provided in the notebook and save it to the /Google Drive/AI/models folder so it can access it when it runs.
Seamless Textures with Stable Diffusion
Generate seamless textures with Stable Diffusion using in-painting available via this notebook. Simply provide the prompt and this specialised notebook can be create seamless tiled images based on your prompt.
Stable Diffusion WebUI – Altryne
Stable Diffusion WebUI allows you to move away from code and provides an easy to use User Interface (UI) that has many feature you may not find in a single colab. Once you run the Colab, you will be presented with a URL like: https://12692.gradio.app
Click on the URL to launch the Stable Diffusion WebUI.
You can create Text-to-Image and Image-to-Image as well as GFPGAN (used to correct and enhance faces generated by AI) and RealESRGAN (used to scale up images in size)
Within the Text-to-Image area (as pictured below) you can specify the size of the image using the sliders on the left hand side, change the guidance scale (I usually keep it at 14 for my experiments), define batch count and size. On the right hand side, you have Sampling Steps – default 50 is usually enough to get decent images.