This is an experimental LoRA trained on AnythingV5 with only 1 image, 100 steps.

The objective of this is to help generate images that resemble the character Oomura Yumi from Chobits. Despite the extremely limited dataset, it succeeds recreating the character which is just what I was looking for, while also keeping the file size under 10mb.

Basic prompt: Solo, 1girl, short hair, brown hair and eyes, red sweater, pleated skirt, blue skirt, OomuraYumi,

It should work very well at any weight between 0.7 and 1.0

Really appreciate any tips to improve.