This LoRA was trained based on Season 3 of the anime.

Trigger Words: Priscilla


test checkpoint = Nlxl

I generated a sample image using the above checkpoint.

The WAI checkpoint was not used because it interferes too strongly with the artwork style.


The results start to deviate from the original starting from a LoRA weight of 0.6.
I recommend staying within the 0.7 to 1.0 range.

Best = 1.0 weight