(November 20, 2022, 8:21 am)GravidaSomnia DawnAI looks okay at a glance:





The lexica.ai site has a useable mobile browser interface, but the way it upscales works better for artistic creations than for photorealistic ones.

ignore the lexica recommendation, they've added censorship back in. An understandable decision, but in practice the nsfw filter is annoyingly overzealous.

mage.space is better anyways.

I just want to put this out here. I use StaryAI which I think is easy enough to take full advantage of. There is not any fine tinkering though involved. But I was able to come up with a few different images. The prompt I used was a simple one, but this is the type of quality you can expect. It isn't perfect by any stretch of the imagination but it's decent. The Prompt I used was "Pregnant BBW" Which fairly enough was a decent prompt to start with. But make sure if using to include words like nude, in order to be specific in language. Pornographic terms usually associated with nudity  will appear with the person naked but other more vague terms will leave the person with clothed type. This one was relatively easy but can definitely be improved with more specificity of language.

(October 2, 2022, 8:09 pm)EngorgedUterus I get the best results by letting the AI modify clothing or create a belly while telling it to adhere closely to the original image, then put the AI belly onto the fully normal photo of the girl with regular manual masking. It can create body shape and rounded lighting pretty well, so works a lot better as a starting place for a manual morph job than it does for just birthing pregnant girls from scratch.

Apologies for the late reply. But what prompts do you use to tell the AI to adhere closely to the original image? So far are my creations are nightmarish. Using Stary AI and mage.space.

(August 28, 2022, 8:00 am)ambrosia Where else would you get to see what it would look like if women with erogenous wombs got pregnant with quadruplets?

perfection. thanks.

If anyone's into weird faces again, here's "crowd of attractive young pregnant girls in swimsuits" using DiffusionBee.
   

Edit: so I'm now playing with "group of beautiful young overdue pregnant women with long hair and large breasts wearing revealing one piece swimsuits laying on the beach" but yeah, the faces.
   

I installed automatic1111 with gravymix, and got much better results with faces, but its understanding of age and group size seemed slightly different.

I'm now proud to present to you all, the Bogus Big-Bellied Beach Babes!

                                       

                   

                   

                   

                   

                   

These should be groups, but one came out solo. No idea why one woman has a towel over her face.
                               

Upgraded a couple of words in the prompt (attractive -> beautiful, large -> huge)
               

And a couple of bonus non-beach ones.
       

Towels over faces is weird but AI can do weird stuff. With txt2img outputs are always going to be random, some things get missed entirely, other things will double up. Most of the pics I share have been edited. To fix the faces I use img2img to upscale the whole pic and it's gotten to the point that turning face restore off reliably improves the realism. The higher the image height the more realistic faces will look, but bodies start to get weird past the 720px mark.

It's annoying me that I can't replicate your outputs using the pic metadata, I assume that's due to the hardware difference.

I am really happy with my latest blend, but again it's not a model for pregnant content (though I've gotten some great pics out of it). I won't be posting for a while but feel free to try out PhotoSomnia as well.

Hey all, I'm just getting started with this... wondering, what's the best version of Stable Diffusion you would recommend? I've been using the online one from stablediffusionai.org and was wondering if anyone knew what flavor they used. I got some pretty favorable results and want to do it locally on my pc. Any recommendations?

Probably my biggest pain right now is the way the AI refuses to allow the belly in my image gens to extend out of frame. I like to try and make them big, but often they wind up being oblong, like torpedo-shaped. It's been driving me crazy, inpainting can solve that a bit but nothing seems to be working quite as well as I'd like it to. Anyone have any tips?