The amount of effort it requires to create a useable work flow built around a genAI llm as the core instead of just a tool in the work flow is absurd.

It basically takes three times the effort to get good results out of it when compared to doing it the old fashion way.

And both require you to still actually understand art to actually produce good results. You can’t just tell an ai to shit out a picture and have it be good. No more then you can tell an artist to just shit out a picture and have it be good.

Both require you to actually put effort in, go though the process and make mistakes fix them. Go back and forth and test ideas.

And that’s the fundamental problem. GenAI is only good enough to at best do the equivalent of a rushed first draft with flat colors. It’s “good enough” not actually good.

And the sort of people using genai arnt the sort of people who are going to go back and edit or clean up the picture. To test and try different things. Hell they aren’t even going to be the sort that custom tunes their model to actually produce what they want.

They just take what ever generic shit model they can find and run with it.

The difference between a generic model and custom trained one is night and day after all. But again the fucking idiots using gen AI ain’t gonna bother to do that let alone know how.

Prompt jockeys are the least qualified people to use genAI. It’s comical.