|
*furiously resumes carving a wooden figure with big titties*
|
#
?
Feb 6, 2023 18:02
|
|
|
|
| # ? Jun 10, 2024 13:23 |
|
|
Fat-Lip-Sum-41.mp3 posted:*furiously resumes carving a wooden figure with big titties* me using a glow forge to cut out a hundred wooden figures with big titties from a board in the same amount of time
|
|
#
?
Feb 6, 2023 18:04
|
|
|
indigi posted:you don't need to understand style convention or canon to make art. see: the first cave painting There is a canon to them though, that provides clues to their interpretation. Size of animals, orientation, presence of human figures and so on. A more recent example would be Etruscan tomb art and associated finds. We get an understanding of the visual language of their society, as well as possible clues of their social and cultural practices, religious beliefs, through what is depicted and how. There are also two canon. Those that evidence suggests were produced in situ, which is to say done by lamplight in a confined space over a relatively short period of time, and those produced in workshops. Now, the latter unmistakably has a higher level of ornamentation and finish, but the former is interesting too as the conditions of the work determined the style of the outcome - use of colours, lines and so on that the artist was able to make out in the dark as well as techniques limited by space. That’s all valuable and provides context to specific burials, sites and so on as well as the visual evidence of Etruscan society.
|
|
#
?
Feb 6, 2023 18:20
|
|
|
Frosted Flake posted:There is a canon to them though, that provides clues to their interpretation. Size of animals, orientation, presence of human figures and so on. but AI has those clues as well
|
|
#
?
Feb 6, 2023 18:59
|
|
|
indigi posted:you don't need to understand style convention or canon to make art. see: the first cave painting do you really think that was the first
|
|
#
?
Feb 6, 2023 19:01
|
|
|
like if you want to say "art is only the creation of self aware entities" that's totally fair and an argument I agree with but once you try to go deeper than that the case usually falls apart
|
|
#
?
Feb 6, 2023 19:02
|
|
|
Stairmaster posted:do you really think that was the first there was certainly a first, by necessity
|
|
#
?
Feb 6, 2023 19:02
|
|
|
indigi posted:but AI has those clues as well I don't think it does, it's unable to put them together coherently. For example, in depicting Egyptian figures: which were based on a grid, had a fairly static canon for a long period of time, the rules of which are known to historians and extensively documented, which in combination should make it the easiest form for AI to replicate, AI still does not conform to the conventions and therefore does not create authentic Egyptian art.
|
|
#
?
Feb 6, 2023 19:03
|
|
|
Frosted Flake posted:I don't think it does, it's unable to put them together coherently. For example, in depicting Egyptian figures: which were based on a grid, had a fairly static canon for a long period of time, the rules of which are known to historians and extensively documented, which in combination should make it the easiest form for AI to replicate, AI still does not conform to the conventions and therefore does not create authentic Egyptian art. it does put them together coherently though I've seen plenty of coherent AI art. but in any case human artists also regularly fail at synthesizing styles coherently
|
|
#
?
Feb 6, 2023 19:04
|
|
|
So is the argument that traditional art is just as uncreative and uninspired as the people who made AI to generate images
|
|
#
?
Feb 6, 2023 19:05
|
|
|
much if not most art is uncreative and uninspired yeah
|
|
#
?
Feb 6, 2023 19:06
|
|
|
indigi posted:it does put them together coherently though I've seen plenty of coherent AI art. but in any case human artists also regularly fail at synthesizing styles coherently  My point is that this is a fairly basic test for replicating art according to conventions of style and it still does not deliver consistently.
|
|
#
?
Feb 6, 2023 19:06
|
|
|
Futanari Damacy posted:So is the argument that traditional art is just as uncreative and uninspired as the people who made AI to generate images all art is uncreative and uninspired. anyone who knows anything about art knows that the only good art is the hypothetical art I am considering making. everything else is derivative, poo poo, literal garbage
|
|
#
?
Feb 6, 2023 19:06
|
|
|
indigi posted:but AI has those clues as well via pastiche, sometimes, thanks to the human inputs on which it's based more simply put when someone sets out to create a visual work, something is intended to be communicated to someone, which is not reliably replicated in the composite/collage output of a text-to-image model
|
|
#
?
Feb 6, 2023 19:12
|
|
|
webcams for christ posted:via pastiche, sometimes, thanks to the human inputs on which it's based That's a better way to put it, thanks.
|
|
#
?
Feb 6, 2023 19:17
|
|
|
Nobody ever learned to draw because they looked at a tree. They learned to draw because they saw a drawing of a tree. - Robert Greenberg The first drawing of a tree was some guy probably forgetting their old rear end word for 'tree' and then just drew something in the dirt.
|
|
#
?
Feb 6, 2023 19:22
|
|
|
The Art is not in the process of asking the machine to display an image, but in the machine reflecting various bits of Human Art back in a fun house mirror collage. It is sort of the same artistic process that happens looking through a kaleidoscope.
|
|
#
?
Feb 6, 2023 19:36
|
|
|
The real foolishness is in thinking the kaleidoscope can produce those artistic images independent of their actual source 
|
|
#
?
Feb 6, 2023 19:38
|
|
|
Futanari Damacy posted:The real foolishness is in thinking the kaleidoscope can produce those artistic images independent of their actual source
|
|
#
?
Feb 6, 2023 19:41
|
|
|
Frosted Flake posted:
so art is something that can be strictly defined through various rules defined over the millennia by the enlightened art knowers, easy to put in various boxes both figuratively and literally, and anything outside those narrow parameters is thus Not Art and therefore worse than garbage got it, thanks for clearing that up
|
|
#
?
Feb 6, 2023 19:50
|
|
|
Tree Reformat posted:so art is something that can be strictly defined through various rules defined over the millennia by the enlightened art knowers, easy to put in various boxes both figuratively and literally, and anything outside those narrow parameters is thus Not Art and therefore worse than garbage A pastiche of Egyptian art that does not understand and follow the norms of representation, scale, style, perspective and so on, is not Egyptian art, no. All history involves taxonomy, particularly art history which had to use it for dating prior to radiocarbon dating, and often still does.
|
|
#
?
Feb 6, 2023 19:58
|
|
|
It's not that art has an exclusive set of rules but that what is classified as art to be created has changed along with the times, to the point where we can easily identify the geography and chronology of pieces based on their appearance I guess you could say something like a more global post-modernist mindset has led to a wider embrace today of what we could call art, to the point where we recognize all those previous stylistic definitions but are ourselves bound only by the conventions of our choosing. So regardless of whether what you would call art is based on rules or not, all you're doing with AI is defining an even more rigid set of rules that it is required to follow to competently produce images 
|
|
#
?
Feb 6, 2023 20:01
|
|
|
I think Frosted Flake's point is that while the machine is capable of some novel results at this point, it's not doing any real thinking about any of this stuff it's being fed. If it were, it could notice things like the rigid characteristics of the Egyptian symbols it's being asked to reproduce. I don't doubt that you could make one of these things, an AI or a bot or whatever you wanna call it, to do that - but I think you'd still have to do a lot of the thinking for it. You'd have to teach it the bounds of the symbols and how the combinations of them may change the context - if the AI could do that, it would be doing that instead of staticky swirls where you get one or two actual symbols popping out. e: at least that's what I thought he meant idk
|
|
#
?
Feb 6, 2023 20:01
|
|
|
Futanari Damacy posted:The real foolishness is in thinking the kaleidoscope can produce those artistic images independent of their actual source drat that's an excellent analogy. great companion to this tweet I saw earlier today 
|
|
#
?
Feb 6, 2023 20:04
|
|
|
Pepe Silvia Browne posted:I think Frosted Flake's point is that while the machine is capable of some novel results at this point, it's not doing any real thinking about any of this stuff it's being fed. If it were, it could notice things like the rigid characteristics of the Egyptian symbols it's being asked to reproduce. Yes, that's correct. The same would apply to medieval Christian depiction of the Saints, incorporation of nature themes in Art Nouveau etc. etc. The other dimension as you said is that art isn't just pretty pictures, it conveys meaning, and symbolic meaning is not something the AI is able to register. Saint Bartholomew being depicted holding his own flayed skin is important to understanding his depiction, which a human artist would understand and an AI does not. Even if it understands that "Saint Bartholomew = holding skin" as a pastiche, that's not really a symbolic representation, simply copying from human created artworks, as with the kaleidoscope. This is really easy to see with one of the most enduring subjects of art, the Pieta. When you ask Midjourney or DALLE to create the Pieta, whatever the specific style or whatever, it clearly does not understand what it is pasting together in what it believes is the right approximate shape. Something is clearly lost in translation, despite this being one of the most common scenes depicted in western art, with thousands of artists in countless styles creating this scene over hundreds of years. Frosted Flake has issued a correction as of 20:17 on Feb 6, 2023 |
|
#
?
Feb 6, 2023 20:10
|
|
|
Pepe Silvia Browne posted:I think Frosted Flake's point is that while the machine is capable of some novel results at this point, it's not doing any real thinking about any of this stuff it's being fed. If it were, it could notice things like the rigid characteristics of the Egyptian symbols it's being asked to reproduce. We can already make bespoke models trained on specific types of images, and things like ChatGPT and Character.AI both use secondary ai to police the output of the generator AI. It's entirely possible to put those ideas together to create AuthenticEgyptianAI or whatever other style you'd want. You'd probably have to do a lot of specific training on images of each glyph and their exact meaning to get readable writing in there, but I'm not seeing anything that a generative system could never do given enough R&D.
|
|
#
?
Feb 6, 2023 20:11
|
|
|
a generative system could never have faith in the Lord
|
|
#
?
Feb 6, 2023 20:15
|
|
|
none of the tech being discussed ITT is "AI" except in the sense of sales, VC fundraising, and naive tech journalism we're talking about layers of probabilistic models with a handful of guardrails
|
|
#
?
Feb 6, 2023 20:15
|
|
|
I wonder if there's a point in the future where something is developed that creates so much backlash AI research is banned or blacklisted in the same way human cloning is banned
|
|
#
?
Feb 6, 2023 20:17
|
|
|
webcams for christ posted:none of the tech being discussed ITT is "AI" except in the sense of sales, VC fundraising, and naive tech journalism well that's why I said "an AI or a bot or whatever" we're just using AI as a shorthand for that, sorry but that's what it's called now even if it isn't actually AI. I don't think Hoverboards are a good name for those powered wheeled things either.
|
|
#
?
Feb 6, 2023 20:17
|
|
|
Frosted Flake posted:A pastiche of Egyptian art that does not understand and follow the norms of representation, scale, style, perspective and so on, is not Egyptian art, no. It's also not Egyptian art because it's not made by ancient Egyptians, but that doesn't make it not art, it makes it not Egyptian. Tbh I think you're overestimating the accuracy people want out of pastiche even when humans are doing it.
|
|
#
?
Feb 6, 2023 20:18
|
|
|
webcams for christ posted:none of the tech being discussed ITT is "AI" except in the sense of sales, VC fundraising, and naive tech journalism okay. is this meant to imply they therefore have no value (the probability engine part, obviously the tech bro part invalidates all this poo poo by itself)
|
|
#
?
Feb 6, 2023 20:18
|
|
|
reignonyourparade posted:It's also not Egyptian art because it's not made by ancient Egyptians, but that doesn't make it not art, it makes it not Egyptian. I said Egyptian art, but of course the same would apply to anything: Art Deco Tiles, Heraldry, German Woodcuts. If people want pastiche, that's fine, but that's not the same thing as passable art. Tree Reformat posted:okay. is this meant to imply they therefore have no value Value? Whatever, I can't really qualify that. Artistic merit? No.
|
|
#
?
Feb 6, 2023 20:21
|
|
|
Tree Reformat posted:We can already make bespoke models trained on specific types of images, and things like ChatGPT and Character.AI both use secondary ai to police the output of the generator AI. It's entirely possible to put those ideas together to create AuthenticEgyptianAI or whatever other style you'd want. You'd probably have to do a lot of specific training on images of each glyph and their exact meaning to get readable writing in there, but I'm not seeing anything that a generative system could never do given enough R&D. this fine-tuned egyptian-art model will still not "understand" the rules of composition of the egyptian art, as diffusion models are not capable of "understanding" anything. it will just be better at making stuff more visually similar to the egyptian art it's already seen. artificial "intelligence" is not intelligent, people keep trying to attribute far more human aspects to it than it really has. that said, defining art based on understanding and following common rules is extremely dumb. i can't draw a straight line or stick to a grid to save my life but my lovely stick figure horus is still at least theoretically 'art' 
|
|
#
?
Feb 6, 2023 20:22
|
|
|
*throws down a "death of the author" card in a huge yu-gi-ohish flourish*
|
|
#
?
Feb 6, 2023 20:23
|
|
|
Tree Reformat posted:okay. is this meant to imply they therefore have no value no, I just think that "AI" is a functionally meaningless term because it can mean anything from your computer opponents in Mario Kart to bayesian statistics to General Artificial Intelligence. generally this thread, according to the OP, is dedicated to Text-to-Image models and it's clearer to say that
|
|
#
?
Feb 6, 2023 20:24
|
|
|
RPATDO_LAMD posted:this fine-tuned egyptian-art model will still not "understand" the rules of composition of the egyptian art, as diffusion models are not capable of "understanding" anything. it will just be better at making stuff more visually similar to the egyptian art it's already seen. artificial "intelligence" is not intelligent, people keep trying to attribute far more human aspects to it than it really has. I think your stick figure is nice  It's not recognizably Egyptian Art, which is my point, even if it shares a subject.
|
|
#
?
Feb 6, 2023 20:24
|
|
|
I think they are art, but that like most art, they suck
|
|
#
?
Feb 6, 2023 20:24
|
|
|
Justin Tyme posted:*throws down a "death of the author" card in a huge yu-gi-ohish flourish* i honestly think a lot of the backlash from artists that isn't about the material conditions caused by the emergence of this tech is a reaction against the idea that anyone opinions about a work other than the creator's matters at all
|
|
#
?
Feb 6, 2023 20:26
|
|
|
|
| # ? Jun 10, 2024 13:23 |
|
|
can't spell "fart" without "art"
|
|
#
?
Feb 6, 2023 20:26
|
|
