This inspired me to see what kind of steampunk skulls Stable Diffusion would come up with. After a few tries, this one was my favourite. It’s also unedited. The text input was: “Steampunk skull, highly detailed, photo, canon35, dramatic, cogs, steam, valves, intricate”
While I’m not going to abandon photography anytime soon for image synthesis, it’s a nice way to play with concepts and ideas that are impractical to photograph.
Fortunately for artist the text input to capture the subtle nuances you’re after for any particular artwork would be very, very, very long. For run of the mill stuff with no artistic merit it looks likely to become a fast way of making images. Of course it will also, in time, produce new forms of art but that art will have very different logic.
Wow Have no clue as to how you’re both pulling this off. But consider me intrigued.
I’ll bet there’s a whole new rabbit hole I could stumble into if not careful.
In my case the skull was generated with Stability.ai 's Stable Diffusion. My current GPU doesn’t have enough video RAM to run it locally, but since it is open source, there are a number of web frontends set up by other people to play with. E.g. Reddit’s StableDiffusion group has more or less up-to-date lists of them if you want to risk the rabbit hole
Haven’t test the AI’s ability to create seamless textures (my side hobby). Might if it a whirl or a variation of such. Possibly might try a Poincare’ instead. lol