My first (working) LoRA, it's big, it's not that good, but it does the job and it's something to startTrained on AnimagineXL v3.1Trigger word: mtkr (or 1girl, moeta kaoruko, comic girls)Recommended strength: 1.0Dataset: 289 images (included mirrored, region cropped images, removed low quality image, captioned using WD1.4 tagger)Train through 16 epochs, 4 images per batch

Moeta Kaoruko (Comic Girls) - AnimagineXL

Readjust some training parameter and reduce lora size

Files with SHA256: 8f012dc108a485c3dcc6008706dec2c8397cb4bb5134e4f3224c8ac944279573