I’ve seen the rise of AI-generated art becoming more and more “convincing”, sometimes photorealistic, sometimes with adorable manga vibes, sometimes looking like it’s inspired by a precise artist or to the contrary looking unique…
And, damn, that’s impressive. Just one year ago I’d have thrown the same joke as everybody, about the hands, about how the drawings seem hollow and soul-less, but nowadays, seeing as it borrows actual inspiration from actual people, and (huge airquotes) “simply” recombines it, I’m not pessimistic anymore. There will be greatness, eventually.
Which leads me to the question I’m asking myself, as I have very little knowledge of how the whole thing works, save vague general notions. Could I ask for a, well, a clarification, from those of you who are into that?
To be super blunt: is it something that can be done, to feed hundreds of drawings by one precise artist to a piece of AI software, have the software munch out those drawings, and then, bam, it will spit out, on demand, art looking like it’s been drawn by the same artist?
If the answer is “yes”, I’m definitely trying that with Gunma Kisaragi, and asking for all kinds of art, from softcore to hardcore
Of course, I suspect it’s far from being so easy.
Maybe a hundred images isn’t enough, and it needs a thousand times more materials. Or maybe the images have to be perfectly cropped for the software to know what to do with it. Or maybe the images require that a human types a precise description of what’s in each image. Or maybe it demands an incredibly powerful PC and the electricity budget of South Africa. Who knows.
But well, here’s for the question. I could totally go for generating art by my favourite artists like that, if it were doable, huhu.
Thanks if those of you who know the stuff can answer! ^___^
A quick temporary answer: thank you very much for your answers already guys!
It gave me some googling to do, I’ll answer later when I have smarter (no, scratch it, less dumb) answers to make!