follow my pixiv where I test out my lora https://www.pixiv.net/en/users/34413372

v2

the model is good at making her lower part. I also tried new settings this time. I normally go with 1/2 of alpa and dim, this time I tried 1/4. the same dataset of v1, but I added duplicates of same images this time. the promp should be the same.

v1

I stopped bothering with making the clothes accurate to the characters from the anime. the model is very bad at making the lower part of her clothes.

the prompt: meredy76, ears muffs, yellow ribbon, ponytail, pink hair, detached sleeves., 1girl, long hair, breasts, red eyes, thighhighs, cleavage, black thighhighs, strapless dress,

with the cape, just prompt cape.

I used screenshots from the anime that I took, I have around used 140 images. even with that many, the model is not good at her clothes.