There aren't many apps that allow generating images with AI for free, since image generation is computationally expensive even if the models are open source. Colaboratory provides free compute, and there are publicly accessible Colab notebooks that can generate images with AI, but I've never seen one that's easy for non-developers to understand or use. So, I made Friendly Diffusion, a Colab notebook designed to allow absolute beginners with nothing more than a free Google account to unleash their creativity with latent diffusion.
With Friendly Diffusion, no prior knowledge is necessary to understand how to use the notebook to generate images. Users choose an open source model such as Stable Diffusion or Openjourney and they type a prompt, and the notebook generates and displays the requested number of images. This is not just a basic tool, however: I wanted to allow users to configure various settings to take control of the diffusion process, so I wrote the diffusion logic myself rather than using a default Diffusers pipeline, and I presented the settings to the user as simply and clearly as possible with sensible defaults. I also integrated Google Drive to allow users to keep the model for later use, save their images to the cloud, and/or provide existing images to alter rather than generating from scratch.
