Trained on 18 images of Bewitching Nami art @ 100 Steps -- It does not do well with fullbody I suggest using the tag halfbust.

This is my first lora I've trained so be nice T_T

Issues:

Positioning of hat on head

Holding objects wrong

Lack of Facial Expression data

Checkpoint used to train Aniverse - V1.6 - Pruned

tags: hat, witch outfit, staff, glass, cat, a drawing, a painting

Sampler: DPM++ 2M SDE Karras

Steps: 20-50

Clickskip: 2

CFG Scale: 7

Lora Weight: 0.8-1

Steps for img2img: 50-100

Denoise Strength for img2img: 0.2