Hello again all! I've been training a Stable Diffusion embedding (basically a spin-off AI) from the pictures in this thread and some of my own archive of similarly younger teens. It's still baking but I've got some early images!

Figured it'd be better to quarantine the output into this sub-forum rather than post this into the main AI art thread due to the topic matter, and frankly these look more real.

Can you elaborate or provide a link to this? because what you wrote sounds interesting, but incomprehensible

(October 4, 2022, 3:54 pm)BiBeBel Can you elaborate or provide a link to this? because what you wrote sounds interesting, but incomprehensible

Absolutely! Stable Diffusion is an image making AI like Dall-E or Midjourney, but it is open-source so it's less restrictive and you can run it on your computer at home. Here's a Stable Diffusion Install Guide video, it's a pain but worth it if you're interested.

I made a new "idea" in the AI called [TEEN-BELLY-1]. Then I gathered 1,119 pictures of teens (the core set was every image in the Younger Teens thread) in the goal age range, made some guiding text prompts ("a photo of a teenage girl posing with her [TEEN-BELLY-1] that outgrew her clothes" or "a photo of a woman's sexy [TEEN-BELLY-1]" for example) and showed them to the AI.

It's a slow process (I have a cutting-edge 3D modeling workstation and it's still days) but the AI is starting to recognize the difference between the pregnant teens I showed it and its preexisting notion of what a teenager or a woman is. Any time I use the term "TEEN-BELLY-1" in prompts now, if I ask it to make a girl or woman it will make her "more like" the pregnant teens I showed it and less like a "standard woman". The images I posted are a byproduct of this process, and at the end I can give it commands like "a high-quality photo of a young woman with a huge TEEN-BELLY-1 who is salsa dancing" and get quality, accurate output.

You can see the AI thread in the art sub-forum for a good example of what happens if you make those requests without the special training first, it's significantly less authentic/realistic looking.

While I was writing this my AI decided to show off and demonstrate that it now understands pregnancy ass, image attached.

Wow ?

Im glad you're running this yourself. That was my second thought. If you were using some AI online, Im sure they track everything.

My first thought was that you've created a new genre...

And my third thought was how long/at what point do you stop doing it for free.

Also... I saw your title...did you know... there is a site called This person does not exist...where you can generate face portraits of people.

(October 4, 2022, 3:54 pm)BiBeBel Can you elaborate or provide a link to this? because what you wrote sounds interesting, but incomprehensible

There's AIs now that can create ANYTHING via references to other art that already exists. He's training a custom AI model by throwing pregnant teens at it thousands of times. I assume he'll release it eventually and we'll be able to use it.

this is amazing. I want to see more!

I love this stuff

This is a very interesting information

I have tried using it AI tech too, but most time I only got semi-realist art, well it's good enough for me for personal use.