Something I was experimenting with. This is a tribadism pose but set from the point of view of one of the participants. This pose has the benefit of only having to generate one female participant as the POV character is just a faceless pair of breasts which means the AI doesn't go crazy playing mix and match trying to generate two people
The pose was trained on exact Koikatsu posed characters with the output being 1024x1024 images so this is the default resolution to generate at and I don't know the effects of generating at other resolutions. Without using quality tags or LoRA's and just generating a straight forward prompt with the triggers you will probably generate the Koikatsu style image
You shouldn't need to prompt for 2girls as two girls are baked into the LoRA triggers and I don't know the impact of generating using the 2girls tag
(TAGS)
TriPV, tribadism, female pov
Description
Trained on 60 identical Koikatsu posed images where the opposite female is changing every time to make the model understand the POV character/pose must stay the same but the other girl changes