It is a visual style booster LoRA.
Trained on a dataset of 100 images, Captioned with Qwen VL
Ep 07/10