Eye candy Asian boys generated by AI

[siscu58]

I wish they were real and available

[Gaybutton]

If they ever graduate to making realistic androids instead of just AI images, then you could buy one and have it pay for itself from the money you save since you no longer would need to pay the money boys.

Maybe you could buy a few of them and then have a real party . . .

[Moses]

If they ever graduate to making realistic androids instead of just AI images, then you could buy one and have it pay for itself from the money you save since you no longer would need to pay the money boys.

Maybe you could buy a few of them and then have a real party . . .

Oh, I think it is very far in the future. Now science cant even simulate just quite simple impressions in Virtual Reality costumes. Even simple dildos and holes are far from reality, while already have heating, and moving functions, some have pseudo ejaculation and moan sounds (via headphones). Modern f*ck-machines are also mostly very primitive mechanical construction while costs few kilos.

[bkkguy]

If they ever graduate to making realistic androids instead of just AI images, then you could buy one and have it pay for itself from the money you save since you no longer would need to pay the money boys.

if you do ever end up getting one, be sure not to let it talk to any money boys, they may give it ideas that will destroy your premise completely - you could end up having to swipe your credit card down an android's butt crack before being allowed to do anything!

and just imagine how intrusive advertising would be in this scenario!

for a bit of only slightly related reading, Ian McEwan's Machines Like Me is an interesting exploration of a male/female/android love triangle that develops when a couple acquire a new toy.

[christianpfc]

Most of these pictures look blurry or out of focus. Is this on purpose? I have AI pictures of humans that are in focus / not blurry.

How much information do you put in to generate these pictures? In particular, where does the underwear style and color come from?

[Moses]

====Most of these pictures look blurry or out of focus. Is this on purpose? I have AI pictures of humans that are in focus / not blurry.===

It depends on drawing style of model (means current AI, not a "boy"). Some models are trained to make ultrarealistic drawing, while other are trained in cartoon style, some are trained for to draw something in the middle. Most models for safety use are trained will slightly visibility of artificial origin.

[Moses]

How much information do you put in to generate these pictures? In particular, where does the underwear style and color come from?

It depends. There are specifically trained models (for example model for to draw flower or car). For to get a lot of nice flowers or cars you haven't to write a lot of description words, but rest objects will be ugly unless you will very precise in description. And there are "general use" models. They are less specific and may draw everything, but with average quality. Than better is trained model and than more "objects it knows, then it is bigger. Typically good model weight is about 6-18 Gb.

I will give you impression how it looks. Here is screen of such interface with some explanation.

1684664132872.jpg

[Moses]

Everything depends on specific skills of model and how much free will you will give to it, i.e. how strict model must follow your description.

Here are few examples of results generated by different models with totally the same instructions. Every example is set of 9 pictures - 9 attempts to fulfill your commands. You will see - some models are trained for cartoons, some for anime, some for porn some for yaoi, some for general use.

Positive tokens are:

raw full body photo of 18yo Asian male, detailed Asian eyes, smile, smooth Asian skin, short messed hair, detailed face, (freckles:0.8), small nipples, 18yo Asian male, black briefs, gold jewelry, flat breasts, flat chest, full body shot, rear view, garden, realistic, photorealistic, vibrant colors, (cinematic light:0.5), (global light:0.3), <lora:pantybulge_v10:0.3> <lora:blondBoy_:0.1> <lora:prettyBoy_:0.1>

Negative tokens:

girl, loli, woman, facial hairs, abs, nude, naked, pussy, vulva, vagina,

Results:

grid-0000.jpg

grid-0001.jpg

grid-0002.jpg

grid-0003.jpg

grid-0004.jpg

grid-0005.jpg

grid-0006.jpg

grid-0007.jpg

Note: next model for sure is trained for draw porn. Since in my instruction models have a lot of free will, it drew some guys without pants, but "penis" is not in my instruction, so results are quite disappointed from realistic point of view.

grid-0009.jpg

grid-0010.jpg

[Moses]

last 4 examples:

grid-0012.jpg

grid-0014.jpg

grid-0011.jpg

grid-0013.jpg

[Moses]

And one more point for understanding: you may choose start point for to make all attempts for to draw picture. For all (all!) pictures above starting point was this picture, I just gave to model enough of free will:

00027-3478016067.jpg