>Duncan using an AI sloppa logo for his new painting competitionFrankly disgusting.Generative imagery stolen from scraping real images people have spent time creating is an insult to everyone with eyes.
>>94835000AI art sucks and people are peasants but it's for a competition, not the next Bible. What do you think an artist will do for such a small piece? Use AI
>>94835000How can you tell?
>>94835187Pretty obvious once you've seen one AI logoFor example the elbows are different pieces of armour
>>94835000This is a low-tier art job at best. No one worth paying for their skills is losing out to this. As for "stolen" art, no AI actually keeps the art it uses in its training data as further reference. Its still better humans simply tracing artwork and claiming it as their own.Have you ever used canva? There are so many pre-made template for you to use, you don't need to add anything unique beyond the text. Does that count as stealing?
>>94835197I don't think so. Look at the stars above the shoulders, they're symmetrical and robots can't usually get that right. Plus, the shading is consistent.
>>94835197actual plate armor (particularly in the Italian style) commonly did that, so you had more protection on the side facing your enemy while giving your own weapon arm better mobility
>>94835030They'll draw it. It would take twenty minutes, tops, especially since they cheaped out and reflected the fucking thing anyway. How stupid are you?
>>94835000idgafcry about the fact you can pump out easy images now. Cause I know I wont.
Traditional games?
>>94842303This board is overmoderated
>>94835767Holy fuckHow will OP ever recover
>>94835000>ai badI'll keep generating my free images
>>94835463>as further referenceanon it is the only referenceit is literally just a denoiser algorithm run repeatedly, using the stolen (yes, stolen - as in "algorithmic mass-scale violation of copyright by webcrawler databases") as the entirety of it's 'training data.''AI' does not think, does not plan, does not even do any decisionmaking (that's what interpreters do; the model itself just spits out weights).it is just outputting a probability distribution given inputs according to the probability distributions that existed in the training datait is, in essence, a weighted pixel shuffler. the pixels and weights being shuffled are derived exclusively from the training data, which is overwhelmingly non-legally obtainedt.actually works with these fucking things (this is a catastrophic bubble that might make dotcom look tame when it pops - people do not understand how much of a loss the industry is running at to make these things look like they're improving when basically everything has plateaued and they're just throwing more GPUs at the problem; there hasn't been an actual breakthrough in AI since the late 20th century, literally everything you're seeing is scaling a vision processing architecture from the 70s or 80s to obscene size via hardware and just kinda hoping it develops intelligence by literal magic, aka "emergence")
>>94835000goodfuck "artists", I've commissioned stuff multiple times only to either get shit art, or to get decent art and months later find out they traced from somewhere.And no, I wasn't a cheapskate.
>>94847704And now you can generate thousands of pieces of shit art that was traced from other peoples' works! Thank you AI!
>>94835601>they're symmetrical and robots can't usually get that right.Here >>94835197 there's a bunch of other logos made with AI that are completely symmetrical.
AI is nothing but equity in action applied to a skill. This was inevitable.
