Hello Ulthy here. I created my first LoRA ever and decided to share. Trained on 81 images using AnythingV5Ink_v5PrtRE as training model.

I aimed to replicate anime version of this character.

As for generation I like AnythingV5Ink_v5PrtRE and Corneo's 7th Heaven Mix as base the most so far combined with kl-f8-anime2 vae.

Some settings I used for previews:

Euler a and DPM++ SDE Karras

Resolution 512x768, hires. fix, upscaler: RealESRGAN_x4plus_anime_6B, highres steps: 10, upscale x2, denoising strength: 0.4

sampling steps: 20

CFG scale: 6

Clip skip: 2

As for tags at the beginning use:

louise zero default, louise francoise le blanc de la valliere, 1girl, solo

Then for her default looks use:

pink eyes, pink hair, long hair, very long hair, wavy hair, hair between eyes, collared shirt, dress shirt, long sleeves, white shirt, black skirt, black miniskirt, pleated skirt, zettai ryouiki, black thighhighs, mary janes, black footwear, black cape, bolo tie, pentacle, pentagram

Negative tags I recommend:

holding, holding wand, wand, from back (if using cape)

This LoRA works decent with other checkpoints too tho I didn't test it that much, Hope you like. I might update this in near future.