Hopefully a model that can be used in a lot of different circumstances, it was trained on 74 images with very descriptive captions. Can deliver realistic images between 20-40 steps on most samplers.