Faithfully recreates Hermione Granger with strong facial recognition, expressive brown eyes, and her iconic hairstyle. Trained for consistent results across Hogwarts scenes, portraits, casual outfits, and alternate universe settings. Works well with anime, illustration, and semi-realistic checkpoints.
Description
This version is still a work in progress and currently lacks stability.
The LoRA can produce good results, but it still has some limitations. The hairstyle tends to be too locked into one specific look, and the face pose can also be too repetitive or inconsistent depending on the prompt.