Back at the end of 2022 I started playing around with DALL-E and Stable Diffusion to create images of Supergirl and other superheroines using simple text prompts. The images that the Ai was able to create were impressive but still pretty basic and rough around the edges.

But in the last couple of weeks… man, some talented people out there are creating some CRAZY images. Hyper-realistic photos of superheroines, but particularly Supergirl. Now, my understanding is that they’re not just opening up Stable Diffusion, typing in “Supergirl” and pressing enter. I think there’s some 3d trickery, some Photoshop refinement, and other magic going on under the hood. But man oh man, the results are incredible.

Check out some of this crazy awesome artwork:

Transformerman

DeviantArt link

Cheddar Design

DeviantArt link

Buffy2ville

DeviantArt link

But is it art?

Over on Deviantart, nyctophobia11 published a post outlining how he and Transformerman created a 3d model and then ran it through an “Ai machine” and came up with these out-of-the-world images of this utterly gorgeous Supergirl. Transformerman has a theme going on Deviantart called “But is it art?” where he is publishing his journey in creating some of the images above. You should check out his profile.

He started with his:

And then ran it through all the Ai supercomputer stuff, and came out with these three:

And then taking the one he liked the most (the underboob, obviously), came up with this beauty:

If you’re interested in reading about the full process, take a look here:

---

It seems like a big leap has occurred in the past couple of months. I don’t know whether Ai technology has gotten better, whether people are creating new tools to complement the Ai technologies, or whether they’re using it in more imaginative ways. Possibly a combination of all three. But it’s exciting because some of the images being created now are stunning, and I can’t wait to see what the artists are going to be creating over the next year.