I am happy to share with you my first artstyle Lora. This is aimed to produce images resembling the style of Asakura Mitsuru.

I think the outcome is decent enough to be published, but i have to disclose that even toh the model was initially aimed for both Illustations and Manga art, it produces much better images when working with manga content.

The Lora is pretty much plug and play.

It works fine at strenght between 0.66 - 1.

You can often leave it at 1, but if you wanna be sure to minimize the possibility of bleeding set it at around 0.85 and you should be fine.

It is trained on a huge amount of images (around 250) and it contains nsfw stuff, so if you are too vague, the outcome will be mostly nsfw (i.e the prompt "masterpiece, best quality, 1girl" might, or might not, generate nsfw stuff).

I am aware of few limitations that comes from the dataset used:

the tag "green hair" when used with monochrome will often generate images in greyscale but with green hair (if you want you can remove the tag, lower it's strenght or boost the monochrome weight to fix this issue).

If you attempt to break the monochromy using tag like "red eyes" (to be more clear if you are trying to produce an image with glowing eyes of 1 color, while keeping the monochromy of the rest of the image) you might produce a "multiple view" image, to fix it you might wanna put multiple view tag in the negative prompt, and tweak the strenght of the tags.

As always you can support me by using this model and sharing your work, by reacting to images or by tipping buzzs.

This model required more work than others, so let me know if you enjoy it.