Trained with 20 Epochs and uses cosplay photos as datasets. This one is bit tricky to train due to how complex its character design, however with right prompts, this LORA can generate (near) accurate image representation of Megurine Luka.

Recommended strength settings: Between 0.4 and 0.6

Do not use default strength settings, the image will be undesirable.

See images here first for prompt examples.

NOTICE: This LORA model is deprecated. Please try the newest version of Megurine Luka LORA but only on SDXL.