https://danbooru.donmai.us/posts?tags=blushyspicy+&z=5

MOSS Image :

Trained with 1k+ dataset, might have some signature or watermark remaining, since everything is AI processed (cleaning, captioning). So make a feedback if it's too much, i'll do a v2 trying to manually curate it.

Anima

Already know the artist style, but has the signature, watermark baked in. Also the style is a bit a mess since it has also very old data.

This lora is trained only from 2026-2024 data, cleaned of every text possible.

You can still use @blushyspicy to make it even stronger (it was not trained into the data).