previously, I found another model for this same artstyle, but I thought there was still a lot of room for improvement, so i decided to make this lora.

trained with around 850 images, collected and selected from the artist's cgs, I ended up being pretty satisfied with the result.

this is model also serves as a test that i wanted try out, since all my previous loras were done using relatively small collections of images. for the sake of experimentation and fun, this time I tried to use as many as I could (keeping diversity and quality in mind).

i highly recommend to use an intensity value between 0.6-0.9. Using a value of 1 is still viable but there's a higher chance things go weird. (PG sample images use a value between 0.7-0.8)

disclaimer: no intention to infright copyright and this model will be retired if the autor wishes so.