Stable diffusion experiments to capture :
- Artists styles;
- "Memory artifacts" caused by duplicate images on LAION;
- Patterns on attempting to create portraits;
- Tiling probabilty;
- How they differ from original artstyle to stable diffusion's final output;
- How "artists" apart from "those making graphical art" outputs graphical assets and styles within stable diffusion (like writers, poets, architects, singers, dancers fashionists );
- Trying to create a replacement for https://f000.backblazeb2.com/file/clip-artists/index.html as good as https://www.urania.ai/top-sd-artists
- All tests using Google Colab Pro and this notebok image : https://colab.research.google.com/github/ddPn08/stable-diffusion-webui-colab/blob/main/automatic1111.ipynb#scrollTo=Y4ebYsPqTrGb with each mosaic / grid of images taking 3 minutes to render.
- The Remote Notebook runs an instance of AUTOMATIC 1111 from github : https://github.com/AUTOMATIC1111/stable-diffusion-webui