artist-zankuro: Original version of the LoRA trained on 512x. Works best at 0.5-0.75 weight
artist-zankuro-new: Trained on a larger dataset at 768x, sharper fine details and had great understanding of zankuro-isms like areola slip, butt crack, and see-through. Works best from 0.75-1.0 weight.
artist-zankuro-new-64: artist-zankuro-new scaled down from 128 to 64 dimensions. Should be about the same but needs to be slightly higher weighted, 1.0 for full style effect.

artist-zankuro-new is my personal preference

EXPERIMENTAL contains locon and loha versions trained on the same dataset as artist-zankuro-new.