|
|
|
|
|
|
by dirtyid
999 days ago
|
|
|
They can probably train an AI to filter based on human appeal. But IMO there's still room for artists with taste and technical talent to manipulate images closer to a curated ideal. Like the current SD photoshop workflows where generated content creates a base. I imagine once the workflow matures, there's going to be more manual imput again, i.e. drafting specific postures / arrangements for controlnet to block out composition before AI fills to 90%, and then human taste tries to refine/reiterate the last 10%. |
|
|