Thank you very much!

When I train a model, I have six datasets I use: three come from the actual character -- BodyCowboyShot, BodyFullShot, Portrait; three come from datasets I've already established outside of the character as close-up crops - Breasts, Vagina, and Additional (think navel, extreme close-ups of eyes, etc). For the Vagina dataset, I have a set group of good images I've made using various "innie" loras or that I've made with Daz (then re-rendered with QIE/Flux2k9b/Krea2). I've found having those close-up datasets helps a lot with Krea learning the anatomy more clearly. Downside is that it makes all of the women have the same genitals, so I had already planned to make a few different datasets like that and use a script to randomize the chosen images used for that lora.

I did try to have close-ups that contained clitoral hoods, but it never came out with a consistent look and tended to just produce bad results. So I oped for innie look first then people can combine with whatever vagina-specific lora they want for their own creations.

@Kerstal It's possible. I'd have to do tests to see how to best do it first, but I don't see why not.