>>13468 The thing is none of the images are actually all that great

I know, but I have seen a couple that turned out okay, plus, the bad results are usually pretty funny to look at.

some of the things I have ordered, although it has a significant amount of imperfections, could serve as "references" for AB/DL themed drawings.

I used the free version of ai maker and this is what I got when I asked it to make me a diaper girl but in an anime version.

I used Neuralblender and this was the best it could do.

I've had the most luck with NeuralBlender out of anything. Prompts in the file names.

>>13717 >mtn dew diapers Lol, please do same but with keyword "mlg"

by editing one of the results to make it as symmetrical as possible, a certain silhouette was formed in the centre of the cloth.

>>13926 It's like some sport team logo

some pretty interesting results

I've been getting some halfway decent results using "__ cartoon in diapers" as a prompt in Craiyon. Some are even good enough that you can trace over them, for less nightmarish results.

some celebrities generations

I got this just by typing "anime girl wearing nappies".

>>14361 >>14358 >>13472 aw dude thalidomide victims in diapers are so hot

jeez all of these look like shit. maybe wait a few more years until the AI learns how to draw fetish stuff first. First you have low tier sissy shit, then furry garbage, then this.

>>14602 The prompt, I'm dead

Did a few myself, anyone got any requests?

>>14847 >got any requests? stop

Stop with this shit, they look terrible. Are you so content starved that a shitty illustration resulted from combining computer generated mess with nightmare visions gets you going? >Technological curiosity/innovations? Yes, I'll get behind it. >It creates good content? NO

>>14848 >See these >oh those look kinda cute for once >click them >OMFG burn them with fire why god!

Yeah, the Craiyon AI is bad at faces. With the right prompts it can do the rest fine. Here's an example.

>>13677 Update, I passed that drawing to Wombo ai used different results so that I could give it symmetry and extracted the “best pieces” from each variant of that result to form this

Perhaps if we try to take the best parts of each result and then resort to forming one symmetrical enough to be recognisable and familiar to a character in nappies or a nappy as such and that same sledgehammer we unify it with tencent arc and then use wombo ai to generate a result even closer to what we are looking for to only give it further details and touches necessary to have as close as possible to a drawing or abdl image.

>>17238 im just waiting on some new PC parts and im going to try this with my new gaming build. stable diffusion (its a Dalle competitor ) allows a local build and with that you can also do a in-painting and out-painting as well as iterative image to image. so i look forward to feeding it images and only allowng it to correct the errors to see what the AI can spit out, im cautiously optimistic it will be good, ill keep the thread posted.