@Shiebs said in Non-Disney animation thread:
There’s not a general animation thread so I’ll just ask it here, does ai art
First of all, there's AI assisted art, and then the broad term "AI art" that's being used as a catch-all for the harvesting machines. AI assisted stuff is fine, we've had that for years in things like spellcheck and photoshop and video games. Spiderverse created and trained a program to get what looked like inked lines against a CG model. There's nothing wrong with "AI" in general.
The problem is the plagiarism machines.
take any talent or skill to do?
It takes zero talent.
And lots and lots and lots of theft.
Like, stealing millions of images without permission lots.
How much time, effort and talent does it take to make this
That particular image is probably not raw AI. AI doesn't do dynamic angles and dramatic perspectives like that on its own, they had to get a base to feed into it and a ton of reference shots of Luffy.
And if it’s none, what does that mean for the animation industry?
AI assisted stuff is already being used. Like I said it's being used in SPiderverse to get ink lines, and it's being used in One Piece to take traditional pencil drawings and get that specific look it's had since Wano.
AI theft programs will be useless. Because it's dogshit at consistency because it doesn't actually know what it's making. So if you render the same shot 10 times it'll use the exact same basic pose and lighting every time but the background and what they're wearing will be different everytime and there will be a bunch of squiggles and it looks messy.
If you tell it to render Korra? It can do that.
Do real artists look down on this because of its lack of skill, or do a lot of people see it as the next step to making animation faster and less time and money consuming?
Artists look down on it because of the theft.
If you wanted to ethically train a program to make a movie or something, no problem, but you have to provide it with the hundreds of images yourself. It's just a tool at that point.
Also, while that one individual image looks pretty good, AI has a very difficult time going outside of a few stock poses because it falls apart fast if it tries to do anything fancy. If you want to tell a story or get a certain emotion or facial expression or bit of unique body language? It has a very very hard time with that.
Like just as a few terrifying examples.
The AI knocks how to render a woman. it has no idea what to do with that stylized face or to put it onto unusual proportions, and the face and body are rendered completely differently. Also the hands are really messed up.
Also, it will use the exact same tricks EVERY time because it is taught "this is what looks good" and then never varies it at all. . You need a human hand to guide it.
Like just ignoring the awful foot here, notice how the lighting shadow is hitting the side of the face and then casting a neck shadow over the torso to frame her chest? That's pretty cool lighting!
Individually they look okay, good even . Its a very dramatic and effective lighting choice and that messy hair looks like flowing motion and those backgrounds look cool. But once you see a few of them they look very samey and you start to see the lack of soul in it. Hard to quantify in a single image, but once you've seen a hundred of these and they ALL have the same poses and lighting patterns it starts to look bad, even to an untrained eye.
And you can see the nonsense in the outfits more clearly here.