If you pay attention to these sorts of things, you might know that MIdjourney is in an open beta program now. What is it? It's a web app that lets you turn text or image descriptions into images.
TL;DR: the full album is here: https://photos.lewman.com/Explorations-with-Midjourney/
Here's an example, "Marilyn Monroe in a cocktail dress" results in:
Of course, you can pretty much create whatever you want, within reason.
A downside to Midjourney is they have a secret list of banned words. I only found this out after reading the Terms of Service in full, and running into some of these words while trying to generate some images. For some reason, the word "censor" is banned, so I can't even make an image about protesting censorship, because it's banned. Ironic? And by pushing too much, I was blocked from the service for a short time for trying to use the word to generate images.
I fed it a high quality selfie and styled it as a White Zombie picture.
I've also worked on photo realistic compositions:
Still a work in progress, but it's pretty impressive so far.
It really struggles with cars, wheels, and animals. It can't quite get them correct. More that it gets the idea, but not the details.
I mean, the general idea is there, but the execution is weak. Even when priming it with a full resolution picture of a Ferrari GTC4Lusso, it wasn't any better. It's the correct body shape, mostly. Mostly the right style lights and grille. The wheels are a mess.
Getting abstract, one of my favorites is this creation:
None of it make sense, but it all just works, at least for me.
Enjoy the album.