Was originally for a bounty, but op posted an expanded dataset after I started training and I didnt check. Don't feel it is worth re-training mine since another user already made one using that dataset, but figured I'd post my version anyway since I made it...
Basic appearance triggers are "yuzuhaLee, black hair, (black eyes, gradient eyes, ringed pupils), double bun, hair rings, short hair, " after that can add "black gloves, fingerless gloves, china dress" etc for appearance.
UPDATE:
Added the requested SD1.5 version of the original pony lora with a slightly cleaned up dataset and tagging.
I do not have much experience with SD1.5 so while it seems to work ok with my limited testing, there might be things wrong with it that I have not noticed earlier. Hands and full body shots in general are terribad, but from what I understand that is just how it is with 1.5…
Basic prompt is:
yuzuhaLee, black gloves, fingerless gloves, bare shoulders, pants, black footwear, twin braids, double bun, black pants, hair rings, red eyeliner, red dress, china dress
Remove and swap whatever clothing tags according to your needs.
Can also try adding “fighting stance” for combat poses, but expect a lot of messed up anatomy.