weight 0.7~0.8 is best.
1girl, hair over eyes, heterochromia, blue eyes, yellow eyes, solo, frankenstein's monster \(fate\), mechanical horns, veil, elbow gloves, white dress, long dress, white gloves, short hair, pink hair, bare shoulders, medium breasts, headgear
This was the hardest LoRA I've ever made.
The reason was that the dataset had too many different images of her frontal hair. One with both eyes covered, one with only one eye, and so on.
I could use the prompt example and cover both of her eyes with hair, but the difference in image quality from checkpoint to checkpoint is too great to write about in the description.