/ic/ BTFO

>>2843330
it's not that hard to paint like rembrandt anon. any artist in his late 50s early 60s worth his salt should be able to paint exactly like rembrandt.

>>2843339
a human didn't paint it.
a computer did it.

>>2843330
>very limited rembrandt filter

This thread might get 25 replies at best, 50 if lucky

>it took an advanced learning algorithm to figure out Rembrandt liked painting portraits of bearded white dudes with hats on
>Just copy and pasted all of Rembrandt's paintings and mushed them together to get each feature
>Didn't have any artists on to tell them what important qualities of Rembrandt's art they should be taking note of, instead they just analyzed surface level stuff like the shape of the features and basic proportions
>ending painting looks blurry af
Literally anyone could do this with Photoshop.

>>2843354
what if they write a version that will allow for a human to take the creative decisions (proportions, perspective, composition) while the computer do the mechanical part (rendering, brush strokes)?

>>2843358
It could make more appealing art since it's being used as a tool but the stuff they showed still looked blurry and too smooth. They have a long way to go in terms of in terms of getting the computer to paint like an actual painter.

>>2843363
bro, I'm another /ic/ retard, but I will never be able to paint like a pro, just average mediocre tier.

I picked software engineering because is a better major than fucking fine arts.

was wondering if is possible to write a tool that automates the mechanical parts of painting and simply obey the creative decisions of a human.

like an advanced photoshop.

>>2843358
>>2843366
there are not "mechanical parts" in painting, at least not to get those results, you'd know this if you actually picked a real brush instead of spewing /ic/ memes

>>2843382
>what is rendering

>>2843366
nope. It'll be very difficult. I'm not a CS guy but I'm pretty sure they printed this thing out was like they'd print any other image out of a printer, only they also added height variability to it to try to get some depth in there. However, it still worked like a traditional printer not like a painter. It had no concern for brush economy, or direction, etc. There may be stuff out there that can make paintings of whatever random thing you want but it doesn't put you in the drivers seat. It just takes pre-existing refs and mushes them together. The computer doesn't see things like form or gesture or w/e else and I don't think there's been any research done on this stuff at all just cause there's so little demand for it. So imo there's still a real long way to go to get the computer to actually paint stuff instead of printing out stuff. Kinda like those people who learn how to copy celebrity photos but don't understand 3-dimensional drawing.

Also when this stuff does reach that level, I'm sure it'll be a useful tool for artists. Kinda like photoshop for things like graphic design. Even a toddler could use PS and whip up some sort of half assed design but it takes real knowledge of stuff like color theory and composition to create appealing designs. So even when the tech advances, you're still gonna have to have a pro level understanding to get the most out of it.

T. Autistic Retard

>>2843388
I'm planning on getting some painting class in college.

I do think the current methods are simply shit and not how a human makes a painting.

maybe is because they're engineers and not artists.

But for me painting is 99% mechanical.

>>2843391
It's mechanical but there's like a hundred million ways you could approach those mechanics. Some painters just 2 quick little brush strokes and that implies the whole eye. A program has no concept of how that works or how to exaggerate things. Obviously you said you wanna handle the creative stuff but I don't think there's anything out like that even in a rudimentary form. All the current stuff takes reference images and puts them together or uses some sort of paint filter. But even if the type of thing you're talking about did come out,I don't know how well it would perform. Like how they have auto-tweening for animation now but it still doesn't look right unless it's done by hand.

But yeah, it'd be dope af if something like that existed. The art field would probably get 99999x more competitive too cause it'd seem so easy. Good luck with your research tho, maybe something cool will come of it.

Yes one day AI robots will BTFO humans at everything. What are we supposed to do, kill ourselves?
>not merging with them

>>2843366
>was wondering if is possible to write a tool that automates the mechanical parts of painting and simply obey the creative decisions of a human.

Such a machine would need an enormous amount of computing power because at that point, it's almost human and currently, scientists have only been able to simulate human brain activity for a second.

In theory it's very much possible but practically, it will take a while for such a program to exist and be able to understand how a human thinks.

>>2843330
Literally none of this "BTFO's" anything.
It made a portrait that one could do with a photograph or photoshop.

Unless I see an A.I making a technically strong imaginative image, we'll still be here and even when that day comes, we'll still create for the pleasure of it.

>>2843330
>replacing artists with computers
>innovation

wew lads

>>2844146
I doubt anyone is going to let that capacity happen.

I think that at the very least the USA is scared from intelligent/complex AIs

>>2844323
Programmers are trying though. Cause I guess they need more jobs.