My first (working) LoRA, it's big, it's not that good, but it does the job and it's something to start

Trained on AnimagineXL v3.1

Trigger word: mtkr (or 1girl, moeta kaoruko, comic girls)

Recommended strength: 1.0

Dataset: 289 images (included mirrored, region cropped images, removed low quality image, captioned using WD1.4 tagger)

Train through 16 epochs, 4 images per batch