Warning: this gallery contains distasteful themes and AI slop, you will not see a perfect mirror of the anime, be it portrayals or proportions. If you are offended by anything, just leave. I won't entertain it.

I've been using Stable Diffusion Web UI Forge for a few years. I trained many Character LoRA variants of Sango (from Inuyasha: The Final Act), both training and generation got out of hand. Horny and curious rabbit went down the rabbit hole. I am sharing handpicked generations that I've liked or thought have passed the mark. Several models and checkpoints were used across these images. They are all unedited and do not contain dumb watermarks, so you should be able to look up their metadata for referencing.

OK, but, with this many files, there should be a torrent, dude.

Ideally, you could do a similar set with canonical proportions using AI....

Base +7, RedJS -6

I've never made a torrent before but I uploaded one just now. That should help in that regard.

As for your suggestion.. although I would like to be able to mirror the dataset, I do not like the canonical proportions, anime and or the retro artstyle design. I personally like Prefect Pony's style, but I don't really like the bias it has for thick thighs and what not. I'm trying to find the right balance, because I also find the ultra booba mature woman stuff gross. I need to experiment more with Illustrious XL and anime checkpoints personally, and then.. I can probably try. Make doujins too.

Last edited on 23 September 2025, 17:41.

Base +3