THIS WAS MADE WITH DATASET FROM THE OTHER SDXL BULGE LORA MADE BY WIGFISH YOU CAN FIND IT HERE: https://civitai.com/models/223231/bulgexl

slight correction: just realized i actually didnt tag anything thong so dont use that, just use "briefs" "brief underwear" etc. Thons may work if the model already knows what they are though. Using a thong lora may help.

I trained it because I was interested how it would turn out personally im satisfied whether it is better or not than the previous may be subjective. The best strength is 0.65. I recommend lowering cfg some. The trigger is "b41g3" I recommend putting that at the beginning of the prompt. The other trigger is "bulge". I'm aware Sometimes you may get unwanted closeups I suggest putting "closeup" in negative and/or describe the face in the positive prompt. It may just be because im putting bulge at the beginning of my prompt. The resolution may or not make a difference also. You can also add "full body" in positive. Maybe even "head out of frame" or "cropped" in negative with weights. I haven't done much testing yet. Works best with Juggernaut and Amalgamation, maybe Himeros with some adjusting(had problems with it making a ?????). You can describe "brief underwear". Doesn't work with pants I may fix that in the future.