Type | |
Stats | 104 549 |
Reviews | (24) |
Published | Dec 18, 2024 |
Base Model | |
Training | Steps: 1,597 Epochs: 10 |
Usage Tips | Clip Skip: 1 Strength: 0.8 |
Trigger Words | ss-YPC$tyl3, |
Hash | AutoV2 C5BED35440 |
The LORA based on style of artist YuPaChu (Yu Pa Chu) who doing exclusively 2d pinup art of humanoid women with specific (bit anime-ish) faces. Dataset contains mostly newer semi-realistic works.
Despite everything above this model are very good with, let's say, men's fluids because some of pinup variations includes them.
Main trigger word - ss-YPC$tyl3
Sometimes it produces women with bulging nipples, bigger breasts or clothing cutouts so negative it if you want some family friendly content. Sometimes they can also be muscular/athlethic.
Works surprisingly well on higher weights even with character LORAs (considering you lowered char's lora weight). Most of my tests are did with 0.9 weight when used with other loras and 1 when used by itself. Do some test and see what is better for your setup. Weight 0.7-0.8 would be a safe start.
Tested exclusively on Autism_confetti and it works extremelly well with it. Non-realistic Pony based checkpoints should be fine in general.
Missed some sneakier watermarks unintentionally. They are not that bad anyways - small signs. All noticable "all over the screen socials" are removed anyways.
Things turned out to be very well. Love the results this model produces.
Feedback, suggestions and your awesome works are always welcome! Sharing with "Add post" below giving me buzz so i can do more stuff no one will use ;)