|
Baronash posted:Out of curiosity, how do they define generative art? I'm just thinking of the types of tricks that I've learned from folks who do renderings in Blender, where modifiers, complex procedural textures, and particle systems are all used to generate a scene. If I use a displacement modifier to create a mountain, a particle system to create grass, and various shaders and modifiers to add fog, is that generative art? So from the get go "3D renders" are already not accepted outside of specific exhibitions focusing on things like that (They had a "Explorations" series last year which were basically empty game levels/walking sims that people could explore, which was done in conjunction with a local University's 3D/Game Design program). But those are rare and not the focus on the art community which really is 99% focused on traditional art styles and mediums. For digital art in their monthly "open" exhibitions the rules are: 1) It must be printed on artistic supports* 2) It must be framed and hung with gallery wire and 3) the final file for the art must be submitted with the program and version listed along with all layers un-merged for review. It's less of a "here's a vague or as precise as we could definition of 'generative art' that we hope will catch it" and more "We're going to review it piece by piece and if it doesn't pass the smell test we're not accepting it". Another thing to mention is that all of their exhibitions, beyond some special solo exhibits, are done "Salon de Art" style where every piece is up for sale. So there is a desire, if not a need, to avoid pieces that are "fungible". A print out of digital piece of art is already pushing the definition and those usually get capped at like $100 for large pieces. Linocut, woodblock prints, and the like would also venture into this territory IMO but they have long since jumped the border into fine art historically (and technically each print will have minor variations since they need to be hand made). Generally the vibe seems to be "did you consciously make a decision on the how you want your piece to look, and personally, physically, direct your skill and experience to bring that into reality?". Which you can argue that knowing how to use nodes and material shaders in blender counts under, but you can also say that doesn't because you didn't actually draw/paint/model the god rays or shadows or volumetric fog yourself. Which is generally why they don't engage with 3D rendered art as it's simply too different from traditional art that they focus on. It's the same reason that 3D printing is not allowed for the sculpture, ceramics, and jewelry exhibitions. A large part of the art isn't simply in the final "image" or "object", it's about the full process. Which is also why I'm really not concerned with AI art, even though it produces some interesting images. * Support in this context means the surface that the art itself rests on. Paper, Linen, Canvass, Hardboard, etc are "supports". Much like medium refers to the kind of and formulation of the material used to make up the image on the support (Acrylic (hard body/soft body), Oils (thinned, impasto, etc), Tempera, Watercolors, etc). Crain fucked around with this message at 15:28 on Jun 3, 2022 |
#
?
Jun 3, 2022 14:45
|
|
|
|
| # ? Jun 7, 2024 20:29 |
|
|
Leon Sumbitches posted:I found Ministry for the Future incredibly optimistic, tbh, once you get past the first 25%. Unless that's what you're also saying, I can't really tell. It is yeah, and the book lost me after that first quarter, or maybe the author just lost the plot and i was just along the ride. I'm not anti-optimism or anything but rather just the "well I'm sure technology alone will save us" variant that is so common strains my eyes with the rolling. If I had to put money on "crypto will save us actually" or "vampires will eat us all actually" I feel like the more likely of the two is obvious
|
|
#
?
Jun 3, 2022 15:23
|
|
|
Honestly.....outside of the example images from Dall-e themselves, and a handful of people curating the "best" pictures (which still seems to be mostly limited to the same best case scenario images given by Dall-e), the images in that thread of stuff look terrible. They're still amazingly fascinating, but it seems like you're stuck rolling the dice a TON right now to get a decent image that actually gets you something you want beyond "the things you mentioned are in the picture". Granted that's the nature of these structures so clearly they will improve as time goes on. Assuming that the best images aren't faked in the first place.
|
|
#
?
Jun 3, 2022 15:47
|
|
|
Alctel posted:It's depressing as absolute gently caress though and the sequel even more so I think I've seen this sentiment a few times and now that I've finally read the thing...why do people find it so depressing? I guess the ending is sort of a downer by typical media standards but the reactions I've seen are sort of disproportionate relative to other bleak fiction. Maybe the consciousness stuff gets to people who weren't hashing through the same conversation in grad school? I dunno. I mostly saw a fictional setting where humanity doesn't manage to wipe out all multicellular life, people we currently categorize as mentally ill can be valued members of society, there's a niche and actual job title for a thing I'm pretty good at but has limited marketability right now, etc. It's not utopian, but it looks pretty sunny compared to the current trajectory of reality. e: I guess this all just applies to the first book, didn't know there was a sequel.
|
|
#
?
Jun 3, 2022 17:12
|
|
|
Crain posted:Honestly.....outside of the example images from Dall-e themselves, and a handful of people curating the "best" pictures (which still seems to be mostly limited to the same best case scenario images given by Dall-e), the images in that thread of stuff look terrible. Yeah, it's really more of an idle curiosity at this point. The best images look "interesting" but interesting is different from "good." If I'm being completely honest, I think most of the people who think this is some amazing revelation that's going to revolutionize the art world in some way... don't know gently caress all about art. So the ProcGen thing can make the Eiffel tower landing on the moon or an astronaut riding a horse, perhaps even in the vaguely recognizable style of a famous artist. Great. So could pretty much any decent artist, it's just that they don't because it's stupid and useless and no one actually wants that outside of a novelty gimmick. It's like the GPT text generators. "Wow that sure is some minimally coherent text about that thing I said!" It's impressive in a sense, but it's not actually useful on any level.
|
|
#
?
Jun 3, 2022 21:37
|
|
|
PT6A posted:Yeah, it's really more of an idle curiosity at this point. The best images look "interesting" but interesting is different from "good." Yeah, the "oh I guess AI is going to take over human creativity before it takes over labor" takes are banal at best and sad at worst. Human creativity isn't limited to "generates new image in style of X". It's also something that's stuck with whatever it's been fed. It's same blindspot that other ML and AI programmers run into with Bias and Bigotry in the training sets. Being fundamentally unfamiliar with the core concepts of something like "Capital A, Art" and boiling it down to a surface reading of a handful of digital art websites and maybe 3 famous artists* with distinct enough styles to be mostly recognized by the general population at a glance does not inspire fear at the "death" of human made art. Really this will become a thing that is utilized by digital artists to make interesting stuff in completely novel ways, which will be interesting to see develop. The next best thing it'll be good for is non-artistically trained people being able to, probably mostly, produce visual ideas they have without needing to pay a lot of money for an artist to create. However, ultimately, I think it's going to be relegated to the likes of FlashPlayer, MS Comic Maker, and other "creative toys" that have come out over the years that have a low barrier to entry, but still require effort and desire to create anything. Everyone thinks they have AMAZING ideas until they're told to start actually detailing that singular flash of a concept their brain had. Creating anything of value takes time and effort to develop the neural pathways to translate thought into product, regardless of the method of delivery. *Also side complaint, it's always "Van Gogh", "Monet", or either "Dali" or something like "Georges Seurat" where they don't actually know his name but only a single painting they did that's vaguely in the public consciousness. And they are always the most annoyingly facile interpretations of their work and style. Van Gogh gets reduced to "swirly paint" when his actual work is extremely deep. And that's a pun because most of his paintings have literal depth to them that isn't revealed in most pictures. "Starry Night" is basically an Impasto painting as the night sky has thick globs of paint being almost sculpted into place more than painted. Similarly Monet gets reduced to "fuzzy" or "blurry" when a lot of his work is more accurately compared to a layering process which creates this see through effect with transparent paints placed over thick dark sections, paint piled onto of paint creating something that ISN'T fuzzy or blurry but brings out the light of the scene being depicted more than the image the scene is nominally of. Again, lots of surprising physical depth to produce literal shadows in the art that is removed from Gallery quality images. It's like most of these guys only go so far as to google "famous painters" and grab a handful of getty images with the watermark still in them.
|
|
#
?
Jun 3, 2022 22:17
|
|
|
Crain posted:Yeah, the "oh I guess AI is going to take over human creativity before it takes over labor" takes are banal at best and sad at worst. Human creativity isn't limited to "generates new image in style of X". It's also something that's stuck with whatever it's been fed. It's same blindspot that other ML and AI programmers run into with Bias and Bigotry in the training sets. Being fundamentally unfamiliar with the core concepts of something like "Capital A, Art" and boiling it down to a surface reading of a handful of digital art websites and maybe 3 famous artists* with distinct enough styles to be mostly recognized by the general population at a glance does not inspire fear at the "death" of human made art. Really this will become a thing that is utilized by digital artists to make interesting stuff in completely novel ways, which will be interesting to see develop. The next best thing it'll be good for is non-artistically trained people being able to, probably mostly, produce visual ideas they have without needing to pay a lot of money for an artist to create. Yes, exactly. I'd also hazard a guess that most people with these fairly banal opinions have never really seen famous works of art in person, and if they have, have treated the experience of going to an art museum rather like a scavenger hunt (that's certainly the sort of behaviour I see most of the time at art museums, especially the most famous ones). And, as you correctly point out: "in the style of" always gets you some poo poo based on the artist's most famous paintings, which is as likely as not to be not what you want. All the "in the style of Salvador Dali" seems to generate a lot more Persistence of Memory-looking poo poo than "Dream Caused by the Flight of a Bee Around a Pomegranate a Second Before Awakening."
|
|
#
?
Jun 3, 2022 22:40
|
|
|
Just a note on the naming it Dali. I think it's very apt to name this project after Dali. He was a pretty crazy futurist who pushed the envelope on art creation, while still being an undisputed classical master. While the shallowness of tech project naming is very apt and true. They all suck. In this case, if Dali was still around I have no doubt he'd be loving around with procedurally generated and AI nonsense.
|
|
#
?
Jun 3, 2022 22:42
|
|
|
PT6A posted:Yes, exactly. I'd also hazard a guess that most people with these fairly banal opinions have never really seen famous works of art in person, and if they have, have treated the experience of going to an art museum rather like a scavenger hunt (that's certainly the sort of behaviour I see most of the time at art museums, especially the most famous ones). I'm extra annoyed about this because I JUST went to the MoMA over Memorial Day and saw so many amazing (and also IMO poo poo, but famous) works of art. Lots of stuff that I actually didn't even know they had in their public collection. Rothko, Newman, Van Gogh, Monet's water lilies, Henri Matisse (both his paintings and his cutouts), Dali, Kahlo, etc etc. So it's all still very fresh in my head. Also at the art commune I'm starting to hang around I'm already seeing tons of very creative stuff from people who aren't known that is so cool. So much cooler than the stuff the AI is making because it really is novel. People really think that art ended in 1975, then in the 80's and 90's we got the "Jazz Cup", then nothing till Banksy showed up and then we got another round of lovely pop art. It's annoying and sucks and I blame capitalism as with all things. BlueBlazer posted:Just a note on the naming it Dali. Oh 100%. And I really do like the idea of the AI generated art because it's a cool new tool. Same with every other advancement in art with technology like 3D modeling, 3D printing, and hell all the way down to the loving metal paint tube being invented in 1841. Idiots always think THIS new tech will kill traditional art and then a decade later it's nothing more than a new tool that artists use to make even cooler poo poo.
|
|
#
?
Jun 3, 2022 22:54
|
|
|
Crain posted:Yeah, the "oh I guess AI is going to take over human creativity before it takes over labor" takes are banal at best and sad at worst. Human creativity isn't limited to "generates new image in style of X". It's also something that's stuck with whatever it's been fed. It's same blindspot that other ML and AI programmers run into with Bias and Bigotry in the training sets. Being fundamentally unfamiliar with the core concepts of something like "Capital A, Art" and boiling it down to a surface reading of a handful of digital art websites and maybe 3 famous artists* with distinct enough styles to be mostly recognized by the general population at a glance does not inspire fear at the "death" of human made art. Really this will become a thing that is utilized by digital artists to make interesting stuff in completely novel ways, which will be interesting to see develop. The next best thing it'll be good for is non-artistically trained people being able to, probably mostly, produce visual ideas they have without needing to pay a lot of money for an artist to create. We get it, you went to art school. AIs can create Art, sorry if that threatens your authority as an Art-knower.
|
|
#
?
Jun 3, 2022 23:05
|
|
|
Magic Underwear posted:We get it, you went to art school. AIs can create Art, sorry if that threatens your authority as an Art-knower. Lmao. I absolutely did not. I'm a tech writer who did Physics and English in college and work at a Tech Start up. Also great reading comprehension.
|
|
#
?
Jun 3, 2022 23:14
|
|
|
Crain posted:Lmao. I absolutely did not. I'm a tech writer who did Physics and English in college and work at a Tech Start up. Also great reading comprehension. They haven't perfected the GPT-3 shitposter, but it's making significant progress.
|
|
#
?
Jun 3, 2022 23:14
|
|
|
Absurd Alhazred posted:They haven't perfected the GPT-3 shitposter, but it's making significant progress. I should have added a "Turn on you are monitor" in.
|
|
#
?
Jun 3, 2022 23:16
|
|
|
This is basically my only thought when robots, AI and labour come up together:
|
|
#
?
Jun 3, 2022 23:36
|
|
|
Magic Underwear posted:We get it, you went to art school. AIs can create Art, sorry if that threatens your authority as an Art-knower. fascinating! an untamed "modern art? my kid can make that" take in its natural habitat 
|
|
#
?
Jun 4, 2022 00:02
|
|
|
Crain posted:Yeah, the "oh I guess AI is going to take over human creativity before it takes over labor" takes are banal at best and sad at worst. It might not take over the entirety of human creativity and probably never well, but it's increasingly rendering graphic designers and illustrators obsolete. People are just uploading a company name and what they do and generating logos and business cards like it's nothing. Maybe it is. Templates, clip art, stock photos...Hell, a lot of businesses just lift other people's photos and logos and don't even think twice about it. I'm getting old and increasingly obsolete. About to pursue a job working in a casino dealing poker games. It pays well and it sure as poo poo won't be boring I wouldn't think.
|
|
#
?
Jun 4, 2022 00:04
|
|
|
if you think about it, photoshop is just *quickly googles* Magritte
|
|
#
?
Jun 4, 2022 00:05
|
|
|
BiggerBoat posted:I'm getting old and increasingly obsolete. About to pursue a job working in a casino dealing poker games. It pays well and it sure as poo poo won't be boring I wouldn't think. I worked at a casino for 10+ years, and poker players are the absolute worst.
|
|
#
?
Jun 4, 2022 00:08
|
|
|
"This will remove the need for illustrators and graphic designers." Yeah, I'm sure the most iconic branding in the world got where it is because the drawing was real fuckin' difficult. "On a red background, the name of a soda in white cursive script" "A large yellow M" "A red chevron shape on white" Oh boy, the finest artists in the world must have toiled day and night to manage the technical mastery to create those, alright! Or... and I'm just throwing this out there for the sake of completeness, since it's absurd on the face of it: perhaps those designers' skill was not in their ability to draw things very well, but to know very precisely what should be drawn.
|
|
#
?
Jun 4, 2022 00:27
|
|
|
Mr. Fall Down Terror posted:fascinating! an untamed "modern art? my kid can make that" take in its natural habitat Kind of the opposite actually. The whole argument is an expression of conservatism and gatekeeping. Like the people that criticized abstract expressionism because it was new and different and still developing. More things are art than what art snobs say, you're a fool to write AIs off.
|
|
#
?
Jun 4, 2022 00:30
|
|
|
BiggerBoat posted:It might not take over the entirety of human creativity and probably never well, but it's increasingly rendering graphic designers and illustrators obsolete. People are just uploading a company name and what they do and generating logos and business cards like it's nothing. Maybe it is. Templates, clip art, stock photos...Hell, a lot of businesses just lift other people's photos and logos and don't even think twice about it. The hard part of logo design isn't drawing a logo, it's coming up with poo poo like this to please the customer.  
|
|
#
?
Jun 4, 2022 00:34
|
|
|
BiggerBoat posted:It might not take over the entirety of human creativity and probably never well, but it's increasingly rendering graphic designers and illustrators obsolete. People are just uploading a company name and what they do and generating logos and business cards like it's nothing. Maybe it is. Templates, clip art, stock photos...Hell, a lot of businesses just lift other people's photos and logos and don't even think twice about it. This is my point. Fine art will just always be there, but being able to name a scene and style, push a button and get it for free is pretty brutal if you make your living from doing that for money. Like even if you think ehh about the dall-e stuff, it's streets ahead* of the melty dogs of few years back, and its not going to stop getting better. *this is an actual phrase in my culture, do not community me
|
|
#
?
Jun 4, 2022 00:39
|
|
|
AI image generation would eat away at the bottom of the pyramid, not the top. Giant corporations who have tons of money to spend aren't going to use AI and neither will people who want to buy 'fine art', but there's tons of space below that for people who want an image of something and think the AI is cheaper and good enough who might have otherwise hired a person to do it. How big a chunk that takes out of the market for human artists I don't know, but I can't look at the results the AIs are getting right now and think it will be nothing
|
|
#
?
Jun 4, 2022 00:43
|
|
|
Magic Underwear posted:Kind of the opposite actually. The whole argument is an expression of conservatism and gatekeeping. Like the people that criticized abstract expressionism because it was new and different and still developing. More things are art than what art snobs say, you're a fool to write AIs off. No one is writing AI art "off". We are arguing against the idea it will replace human art. I've said a couple times that these look like cool tools to play with that will probably lead to some neat art, but they won't replace art any more than 3D printing is replacing sculptors or ceramic artists. If your main complaint is laughing about "computer toucher doesn't actually understand thing they are trying to replace with code" then I'm not sure you know what thread you're in.
|
|
#
?
Jun 4, 2022 01:00
|
|
|
It'll be stock art. We've already nearly mastered AI generated children's cartoons.
|
|
#
?
Jun 4, 2022 01:13
|
|
|
joe football posted:AI image generation would eat away at the bottom of the pyramid, not the top. Giant corporations who have tons of money to spend aren't going to use AI and neither will people who want to buy 'fine art', but there's tons of space below that for people who want an image of something and think the AI is cheaper and good enough who might have otherwise hired a person to do it. How big a chunk that takes out of the market for human artists I don't know, but I can't look at the results the AIs are getting right now and think it will be nothing As a power user who dabbles in layout and publication, I absolutely will use it to avoid a first stage hiring situation, or place holders while waiting for commissions, or even using it to drive a creative towards a specific look. Give me 15 iterations of "snow covered mountain, line style" saves the creative step, that could take a week of hunting, haggling and hoarding.
|
|
#
?
Jun 4, 2022 01:24
|
|
|
Harold Fjord posted:We've already nearly mastered AI generated children's cartoons. yeah that's not true a lot of the generative stuff still breaks apart at the seams once you go beyond a single image or chunk of text
|
|
#
?
Jun 4, 2022 01:25
|
|
|
shrike82 posted:yeah that's not true Sure, but how long is that going to be true for? Dick Trauma posted:I bring to you the gift of many Kermits.
|
|
#
?
Jun 4, 2022 01:27
|
|
|
"beyond a single image or chunk of text" as in videos or longform text i work in the field, specifically in NLP (text stuff) and i get that generative stuff like dall-e and gpt-3 has a wow factor for the layman but most real world commercial use-cases for AI are still in unsexy discriminative techniques there's a reason why years after GPT-x has rolled out, all of the use-cases are in toy settings, no one's been able to roll out a meaningful "enterprise" product around it shrike82 fucked around with this message at 01:32 on Jun 4, 2022 |
|
#
?
Jun 4, 2022 01:28
|
|
|
shrike82 posted:"beyond a single image or chunk of text" as in videos or longform text yeah, forgot to edit in a response - sorry. i just think if you can't draw the line of where we are now compared to where we've been, into the future a few years, and not see it taking over a lot of stuff at the more journeyman end of artistry you're being a little wilfully blind. doesn't mean there won't be work for artists, but picking the best AI candidate out of 100 and doing some touchups is, well.
|
|
#
?
Jun 4, 2022 01:30
|
|
|
That's a lot of frogs and maybe two or three Kermit's. Though it's cool that it parses Kermit as a frog correctly.
|
|
#
?
Jun 4, 2022 01:32
|
|
|
Crain posted:That's a lot of frogs and maybe two or three Kermit's. i read a good piece on AI observing that people who work in the field are vastly more pessimistic about AI because they can see all the seams, but if you put together all the various places advances are happening we are a lot closer to something you'd call AI than the experts think like, i can mumble into my phone in a kiwi drawl and it will pop up something like this  this is an interesting pov from Kasparov though, who famously said a computer would never beat a human at chess
|
|
#
?
Jun 4, 2022 01:46
|
|
|
Looking forward to skynet being racist based on training data.
|
|
#
?
Jun 4, 2022 01:48
|
|
|
Main Paineframe posted:The hard part of logo design isn't drawing a logo, it's coming up with poo poo like this to please the customer. Oh, believe me I know. I try over and over again to get them to keep it simple and iconic, pointing out the most famous logos in the world (McDonalds, Nike, Mercedes, Apple, Twitter, FB...hell, loving BATMAN) and, to a man, they always want me to add a linear gradient, some drop shadows and a silhouette of their dog or truck or some poo poo. I tell them we can add that later but lets get the (very) basic shape right that communicates everything. Where I live, American flag imagery is real big and that fucker is hard to design around or put type on let me tell you. Lots of eagles, stars, stripes and USA USA USA stuff and lot of outlined text. I honestly don't know how those designers managed to sell that iconic poo poo that works so well and is so basic. A lot of people also either like script fonts (Brush Script font is popular), which tend to be cheesy and ugly and the rest want IMPACT font or Helveltica BLACK. TBH, for the poo poo most of them want, maybe AI algorithms are just fine. BiggerBoat fucked around with this message at 01:54 on Jun 4, 2022 |
|
#
?
Jun 4, 2022 01:51
|
|
|
A lot of it is going to come down to how AI products are marketed. Right now every MBA and C-level wants it for the same reason they fall into Blockchain or crypto traps, regardless of whether or not they can actually use it. Once it becomes ubiquitous the first industry or company to abuse it and flood the public consciousness will taint the idea.
|
|
#
?
Jun 4, 2022 02:05
|
|
|
I think there's some confusion here because when we talk about "art" some people think we are talking about fine art, but the vast majority of art isn't fine art, it's consumer-facing art. Logos and icons and banners and whatnot that don't need to be amazing, they simply need to be functional and eye-pleasing. And there are a lot of people who make those for a living. So if a machine can make functional consumer-facing art, or maybe even concept art of larger projects, then a lot of artists are going to be out of a job. It's really that simple, and there's no need to dive deep into what is and isn't art.
Clarste fucked around with this message at 02:11 on Jun 4, 2022 |
|
#
?
Jun 4, 2022 02:09
|
|
|
Clarste posted:I think there's some confusion here because when we talk about "art" some people think we are talking about fine art, but the vast majority of art isn't fine art, it's consumer-facing art. Logos and icons and banners and whatnot that don't need to be amazing, they simply need to be functional and eye-pleasing. And there are a lot of people who make those for a living. So if a machine can make functional consumer-facing art, or maybe even concept art of larger projects, then a lot of artists are going to be out of a job. It's really that simple, and there's no need to dive deep into what is and isn't art. Exactly.
|
|
#
?
Jun 4, 2022 02:13
|
|
|
I get that, but I'm between Harold's point and the estimation that it'll take out part of the bottom rung of art jobs. I don't think it'll kill them totally off. It'll get integrated into art pipelines as bosses make mistakes in trying to eliminate people. I'm sure utilities will be added to things like Photoshop or dedicated programs, but imagining that will mean some boss will just have an intern churn out what is needed is like saying MS Word means companies don't need writers now. I've seen C-levels use snipping tool to take an image of a chunk of text and paste it into a document and complain the can't edit the words. There is no level of foolproof usability that they won't break, so it's just going to create a new kind of job, or rather just recontextualize the existing job.
|
|
#
?
Jun 4, 2022 02:39
|
|
|
Yeah I think we are on broadly the same page. I also think ai weirdness will become an a e s t h t I c if it isn't already lol
|
|
#
?
Jun 4, 2022 02:42
|
|
|
|
| # ? Jun 7, 2024 20:29 |
|
|
Oh it definitely has, especially with "liminal space" and "backrooms" horror stuff. The twisted faces and nightmare machines are great fodder for stuff like scp.
|
|
#
?
Jun 4, 2022 03:00
|
|