I trained this Lokr to solve the issue of not penetrate deep enough. However, I was too ambitious and attempted to use a single model to tackle every problem, which end up messing the original style a little.

The previews show that this model will heavily affects the image, and occasionally it doesn't penetrate deeply as expected.

Surprisingly, it does help with some other problems, like reverse cowgirl, penetrate depth of perpendicular paizuri and some futa problem.

I sincerely apologize for the poor experience you'll had with this LoRA. I might switch to a different training approach and train a new one.

If you found the penetration is not deep enough, you can try (deep penetration:2) or higher.

The compare previews use random prompts in anima base's gallery.