I figured I’d share this free version of Stable Diffusion that doesn’t have annoying G-rated or anti-NSFW filters in place. It’s fun to play around with although it does have some limitations.
I used the Dreamshaper 6 ai (it’s in the drop down menu for model).
I learned that if you don’t ask it for a particular ethnicity it will definitely draw anime characters. If you don’t ask it for men it will draw women. If you don’t ask it for someone older it will draw teen anime style guys.
It’s good at some things that I tried. It can do muscles and spandex with no major issues.
It struggled with BDSM gear and I don’t think it really understood what a chastity belt or device was. I tried feeding it some names of better-known craftsmen and the best I got was some Boris Vallejo style almost fantasy drawings.
It absolutely struggled with latex and didn’t seem to grasp the kind of full coverage that we’d look for in a dronification scene. It did a better job with wetsuits to be honest.
Those were a couple of the better results out of probably 20+ attempts.
It absolutely couldn’t understand adult diapers even when I fed it a starter image so good luck with more esoteric kinks. It doesn’t seem to be able to grasp concepts. I got a lot of very weird results.
And finally, it’s fantastic at power armor and anything manga-esque so by all means go nuts with that. Wanted to see Captain America in an Iron Man armor? It will do that no problem.
One thing that is kind of neat is that if you don’t vary the description of the guy you describe too much, it will kind of keep drawing the same “character”. With some variation, but it’s generally consistent. I think that comes across quite well with the images I’ve shared. There is a distinct sort of blonde-haired bearded “character” that it kept generating and that was because I kept most of the descriptive prompts the same and varied the clothing/gear that I asked it to depict.
If the ai is struggling with what you’re looking for, it can help to give it an image to work off of (the output won’t be a copy of the reference image).
As you can see, the generated image is really nothing like the reference. But without the reference it just wanted to give me drawings of bodybuilders in latex shorts like this,
I’m sure this version of Stable Diffusion will do a good job with twinks too, I just didn’t bother to try. But without a lot of guidance it wants to default to anime characters for sure.
after playing around for a while, I’ve had some luck with getting the program to make twinks. but unless you’re very specific, it will often default to adding muscle, hence some of these pics. but it definitely works!
Thanks for sharing! I had been looking to try something like this for minute but couldn’t find a site that I liked. Didn’t have too much time to play around with it but I’m definitely going to fool around with it again tonight. It seems like it would be really good at following along with a story by how well it handles incremental changes.
I’ve had some luck creating some cool images with AI Image Generator (free, no sign-up, no limits) ― Perchance
I like you get 3, 6, or 9 images from a prompt and it also will have very similarly looking characters. it does struggle with groups but it goes great glowing eyes, sports gear, and collars if you’re willing to work it at.
I also like the option to choose photos from different decades. I played around with an idea for a mind control cult have existed for many decades. This is what it delivered for the 70s.
I tried playing around with it but didn’t like any of the results. Can you post the prompt you used to get some of the first images you posted (with the medieval-looking metal accessories)?
You can choose different art styles. Scroll down to “fantasy portrait” then all the images will have a medieval look. I used a description about a young shirtless warrior or shirtless warrior if you want the older guys. I added stuff about corrupted by evil magical amulet and stuff like that.
Fun stuff, but definitely needs a bit more skills in fine tuning the wording to get it close to what you want. I played around a bit before I managed to create something close to what I want:
Given what AI can do, I’d love to see an author try to write a story where the image supplies something that can’t easily be described in words. Not an illustration of what it’s the text - that’s superfluous. And I don’t mean sexual activities (although it can be that). What I mean are ideas that would make the writing too wordy and would be better expressed by an image. Anyone up for it?
