Yes, I like loras of the most obscure shit imaginable.

Very limited dataset, requires trial and error. I use the lora at ~0.7.

useful terms to include or put in negatives: cartoon style, short hair, 3d, gameplay