D&D General DALL·E 3 does amazing D&D art

Could you create a playmobil version of a D&D heroes?

And a creature like the bairaur, but with traits of Spanish fighting bull, not really a "manotaur" (AD&D creature) but with a head with horns and some human face traits, not totally bull-head. And a ha'pony (centaurkin, mixture of hafling and pony)?
 

log in or register to remove this ad

Trying some stuff out to create some imagery for my "Dragons Must Die" campaign setting

1698418525116.jpeg

1698418548433.jpeg

1698418571760.jpeg

1698418595335.jpeg

1698418617177.jpeg
 

Some of the photographic style images with broadly smiling people give me the impression that it's swung from the "too-perfect" side of the Uncanny Valley of Teeth all the way over to the "weirdly imperfect" side. Not sure if that's a recent development or what, but it seems to be "trying too hard," so to speak.
 

Some of the photographic style images with broadly smiling people give me the impression that it's swung from the "too-perfect" side of the Uncanny Valley of Teeth all the way over to the "weirdly imperfect" side. Not sure if that's a recent development or what, but it seems to be "trying too hard," so to speak.
Yeah, I get what yoy are laying down here. Thing is, the mathematical algorithm can fake organic patterns extremely well.

The real still sticking point right now is mainly the eyes.
 

Is it just me, or are those pictures getting better really quickly. The hands are starting to look right most of the time, for example.
I dunno. It still struggles to get the number of fingers right. It’s still often off by one. And I still get plenty of instances where the hands are on the wrong way, especially when they’re holding something.

I’ll concede it does a better job with hands in the more photorealistic styles. In the various illustration styles, it still produces some pretty wacky results.

It also still sometimes gives me people with two left (or two right) feet.

And it really still struggles to line stuff up when it passes behind something else in the picture. (Look at the Duchess Zirkbane’s tail for an example.)

And if you want to include multiple people, all bets are off.

So yeah, it might be improving rapidly, but it’s still got a long way to go in my opinion.
 






Remove ads

Top