Trained on a larger dataset. Probably better than first one. Feel free to check both
LoRA does exactly what you think it does. Have fun.
Added PDXL version, didn't try it with Pony though. Works well with Compassmix.
