For example, it is EXTREMELY difficult to say, “Keep the characters, but instead of the background being Paris, make it New York and put a flower in her hair”—before you know it EVERYONE has a flower in their hair.
You spend more time telling AI what NOT to do, than what to do. Something that an artist can whip up in 20 minutes can take a day with AI.
It’s the slow way to get art.
Anyone who thinks it’s fast doesn’t know art (managers are ESP SUCKERS to the siren song).
@feloneouscat @gshevlin It's similar to stock photography.
Those who care still want photography of their actual people, and their actual event.
But a lot of folks are willing to settle for a generic, inexpensive, low-hanging fruit solution; honouring the craft doesn't factor in.
You're right: AI images can't be copyrighted. But then again, they're also largely shielded from copyright claims.
AI is disrupting everything, and not in ways that generally benefit artists, writers, actors, etc.
I suggest anyone who thinks AI will save them money needs to do a serious cost/bendit analysis (and not with phony numbers—the reality is not in the favor of AI).
I use it for guides (no more than a ruler) as I do gaming art. It doesn’t replace anything (the art is still generated by me)—anymore than Adobe’s Substance Painter replaces art.
AI relies on the works by people. When trained by AI works, it turns into garbage. AI doesn’t create. That’s only done by us.
@gshevlin
For me, I can say, “Give me 10 iterations of a Frank Lloyd Wright building”—it doesn’t give me the 3D model I need, but it gives me ideas about how I want something to look.” Frequently the art it generates is crap. For example just saying “Futuristic build with glass and steel in the forest” will generate something that is decidedly NOT futuristic.
The idea that it will replace artists is an employer’s wet dream. Sure, try that.
Try to copyright it. Good luck, sucker!