ประเภท
LORA
เวลาโพสต์
2023-09-21
โมเดลพื้นฐาน
SD 1.5
คำที่กระตุ้น
Krupinski
คัดลอก
การแนะนำเวอร์ชัน
I tested the existing character pack and it did not work well for her so I made this LoRA.
Trained on NAI, 169 images in dataset. About 70% of the images are fanart, the rest are anime screenshots.
Training parameters: 3 repeats, 16 epochs. I didn't test different epochs this time but epoch 16 allows the hat to be on and off with the hat prompt so I figured nothing has been overbaked.
Trigger word is just Krupinski, no need to tag short hair, dark skin, green eyes, light brown hair.
