A simple style LoRA made from a dataset of generated images from my 1.5 models.

Trained on PonyDiffusion v6 and tested on AutismMix_confetti

Works well at weights 0.8 - 1.0
I recommend 0.8