I had almost forgotten this artist's work, but since a user uploaded a character LoRA based on their art, I decided to train a style LoRA for them as well. While testing, I found that the base model captured the original art style more faithfully; however, I chose to upload images generated using the "One obsession" model. I will also upload a separate set of images generated with the base model below for comparison. All training materials were in color; no black-and-white manga assets were used. There isn't much else to add—I will take this down if necessary.

One final note: I used "anima.base" for training, but the generation results were poor and prone to style blending. I recommend using "One obsession_Anim" instead, as that was the primary model used for most of my images.