When we migrated from Pony models to Illustrious ones, we still wanted to generate images with the styles from previous checkpoints/LoRAs that didn't work anymore. So, we trained LoRAs based on a few images, and created these models.

This is part of the TT series of models, LoRAs made with tiny training sets (1-6 images). These models were trained on existing AI generated images from Ponywave