Experimenting with AI Image Generation using Stable Diffusion

December 4, 2022
Brian McBreen, Independant Scholar, USA

Synopsis

The following are selected examples from my experimentation with Stable Diffusion using DiffusionBee on Apple Silicon. Dozens of images were produced for different themes and prompts, occasionally varying the parameters to note the impact. I have included the images captioned with the prompts for the handful of examples below.

By and large, nouns with prompts produced the best images. The algorithm seemed to struggle more with abstract concepts–an area I will continue to experiment with. Most took less than 30 seconds to produce at the default 512×512.



Reference Notes
  1. René Magritte
  2. M.C. Escher
  3. Albrecht Dürer