Works with most anime models. Realistic models might not work, I don't use any.
If you like my LoRA, please consider giving me a rating!
Best weights: 0.6-0.8 - canvas size in portrait formats work well
Use 0.6 if you only care about the general shape of her hair. For results accurate to the characters, use 0.7-0.8.
For best results, include the following tags:
rio, white hair, long hair, blunt bangs, yellow eyes, medium breasts,
For her original outfit, include:
school uniform, black jacket, white shirt, plaid skirt, socks, shoes
Models used for preview images:
MizuMixV5: https://civarchive.com/models/103270?modelVersionId=110783
VampeliumMixV2: https://civarchive.com/models/91829?modelVersionId=113096
XXMix_2.5DV1: https://civarchive.com/models/54416/xxmix25d
Description
Release
FAQ
Comments (2)
Thanks for your dedication. I'm sad there are only few Eternal Return's character's LORA except your works. in light of that I'd like to create other characters LORA. I'm wondering if you Could share some details for your Lora creation process, like number of repeats or epochs, learnig rate for Unet and text encoder. Of course, If you don't mind it.
Oh, actually that's very simple, I used HoloStrawberry's guide! https://civitai.com/articles/4/make-your-own-loras-easy-and-free It's very intuitive and should help you with what you need to know to make your own LoRA! Only collecting Datasets can be very annoying sometimes haha