AI art is just tracing on steroids, and here's why.

[Anonymous No.6639298]

>AI can't understand form, they can't understand it as a concept, they're LLM and that crap is nowhere near AGI or Strong AI.

>LLM is just a confident bullshiter, they have no idea if they're right about anything, but they will present it as if they're are.

>Current AI can't understand concepts, they can just replicate and remix their dataset, just like a next word predictor.

>That's how machine learning and deep learning works, they find patterns in their dataset and try to create something "natural" without overfitting too much aka tracing 1:1

>AI art is just a scam to trick artlets, big corpos and /beg/ artists into giving money to scammers.

>Modern AI is barely improving at this point, they pretty much stagnated, even GPT-4 is shit, and that's because they can't reach human levels, nor they will in the next few years or decades.

>You need an entirely new type of AI to even come close to a human level of understanding of concepts.

>Current AI is closer to a parrot than a human.

Can we finally go back to drawing and stop this AI meme?

[Anonymous No.6639303]

>>6639298
I MADE IT FIRST NO ONE CAN REPLY BELOW THIS WALL
________________________________________________________________________________________________________________________________________________________

[Anonymous No.6639306]

I read somewhere that AI generated stuff is the ultimate expression of nihilism. That if no one can really create meaning it doesn't really exist.

[Anonymous No.6639326]

>>6634700
I can literally tell just by looking at a thumbnail that it's AI if I see one. And since AI exist there will be technology and laws to prevent future models to be trained with future real artwork, making SD and others being stuck with their current artstyles. Also it can never create complex scenes with emotion and meaning in them, you can't even generate the most simple stuff like a girl spilling a cup of coffee after bumping into someone

AI is literally only good for closeup faces and poses. You can create some pretty good stuff though through control net in SD. You could draw a sketch of a complex scene and put it in to get finished in 5 seconds. Lightning, color, everything.

[Anonymous No.6639337]

>>6639326
>>girl spilling a cup of coffee after bumping into someone

I just tried to generate that. It's horrifying.

[Anonymous No.6639355]

Tracing on steroids.

[Anonymous No.6639392]

>>6639298
Ai is basically dress up flash game like you see on picrew. The art is pretty when you see but the moment you learn the image was generated its value plummet to zero. This is why people art so desperate to convince others that ai art is real art. They want the respect as much as high level pros art without the work of one.

[Anonymous No.6639523]

>>6639326
>AI is literally only good for closeup faces and poses.
>poses

I have a friend that uses AI art and he said that he can't get all the poses he want.
He wants those crazy hentai poses.
Lately he been learning how to pose 3D models, and now he finally can get all the poses he wants...

[Anonymous No.6639529]

>>6639355

Is that an AI version of Barret Wallace (FF7) ?

I saw a lot of pictures very similar to your picrel.

[Anonymous No.6639536]

>>6639355

>First pic pose
>Second pic blue lighting
>Other pics muscles, proportions, etc

[Anonymous No.6639598]

>>6639298
COPE: THE THREAD
Kek

[Anonymous No.6639713]

>>6639598
doomer has given up: the post

[Anonymous No.6639799]

You'll never know what I'm putting into my machine. Long as it ain't published, you'll never know anything.

[Anonymous No.6640432]

>>6639523
Show me his work. I know that you can put shit like that into controlnet, haven't tried it myself.

[Anonymous No.6640438]

>>6639298
Recently saw something about how an amateur beat an AI trained to best the the top talents of the game GO! - it'd even beaten the world champion and made him retire. How'd the amateur beat it? By playing so amateurishly the AI didn't know what to do.

These AIs don't know what they're actually "drawing", it has no concept of anatomy, or colour theory, or of anything really - just like how the GO AI has no clue how to actually play GO.

>>6639355
That dude needs to some moisturizer or something, he's got reptile skin.

[Anonymous No.6640442]

>>6639298
Current ai is nothing more than an advanced denoiser.

[Anonymous No.6640468]

>>6640438
>By playing so amateurishly the AI didn't know what to do.
it wasn't that, they found that they ai actually has no concept of groups and designed tactics specifically to exploit that
https://youtu.be/l7tWoPk25yU

[Anonymous No.6640469]

>>6639523
Do you have the full page of that pic?

[Anonymous No.6640470]

>>6640442
Current humans are nothing more than an advanced Morganucodon.

[Anonymous No.6640505]

>>6639298
>>That's how machine learning and deep learning works, they find patterns in their dataset and try to create something "natural" without overfitting too much aka tracing 1:1

this is actually how a real biological brain works

[Anonymous No.6640566]

>>6640468
You can see this in even in the prettiest, most polished AI piece. There is something deeply vapid about them.

[Anonymous No.6640614]

>>6639298

[Anonymous No.6640621]

>>6639298

[Anonymous No.6640630]

>>6639298
LLM is like a game of charades where the program tries to fill in the blank based on your prompt and the dataset. It can't come up with anything but you can kind of trick it into coming up with novel answers. All AI can do is fill in blanks based on the model

[Anonymous No.6640678]

>>6639326
You're fucking deluded. What did AI art look like 5 years ago? What does it look like now? What will it look like in 5, 10, 20 years?

[Anonymous No.6640683]

>>6639392
Nobody is going to pay for art once they catch on that they can make a poster for their event in 2 seconds for free. We live in a darwinian capitalist economy that only rewards being cheap.

[Anonymous No.6640685]

>>6640621
>aifags taking pride in being single digit iq no internal monologue npcs
?

[Anonymous No.6640692]

>>6640630
>All AI can do is fill in blanks based on the model
Check out "emergent behavior". https://duckduckgo.com/?q=AI+emergent+behavior

"Emergent behaviors are complex patterns or behaviors that arise spontaneously from the interactions of simpler elements or systems. They are not explicitly programmed but instead emerge as a result of the underlying system's dynamics".

This might be what is phreaking people out. They think it's just filling in the blanks but it's really coming up with new stuff. A lot of people are having problems with this. Some real cope.

[Anonymous No.6640816]

>>6640614
>>6640621
To start with, this maymay has "Don't Dead Open Inside" tier formatting. It should have started with soijak smugly declaring the artistic merits if le red canvas with chad responding with "It's just red", so that the top and bottom would share a consistent "Here is thing" -> picture of thing -> "This is my response to thing" format.
The main problem however is that even formatting this meemee wouldn't make it any less the intentionally dishonest strawman garbage that it is. You're portraying your detractors as conceptual fartists and MOMAniggers, when those that you're actually in direct conflict with are illustrators and visual artists- digital ones, to be more exact- and are completely diametrically opposed from the bananawalltapers and degenerate paint shitting retards as you can get.
If you're seriously going to argue that if a red canvas or squirting paint out of your ass can be considered art, so too must your algorithmic sloppy seconds be, then you are REALLY setting a low bar for yourselves. Like "bottom of the Marianas Trench" low.

[Anonymous No.6640862]

>>6639298
Can you trace as good as AI?

[Anonymous No.6640864]

>>6640614
>>6640621
AI fags are mainly the ones who support the idea that banana on wall is art, and the banana on wall fags are not going to get impacted by AI. so why would top right and bottom be the same person? retarded pajeet

[Anonymous No.6640974]

AI art, like any piece of art, is art only IF the human observer determines it to be.

Banana on Wall is art because people say it is with more money and louder than people who go 'but it's just a banana on the wall'.

No, no AI artist will ever be able to claim as much respectable experience as someone who makes art from just a pad and a pen, but that doesn't make the AI artist not an artist. Just nowhere near as respectable. Just how making an ikea chair makes you a craftsman, but no one respects you even as close as to the dude who carves chairs out of wood with a chisel.

You're all nerds arguing over semantics while neoliberal copyright hoarders ruin everything. Stop fighting over things that don't matter and just draw if you want /ic/ to respect you.

[Anonymous No.6641121]

>>6640974

"Art" is anything that elicits an emotional response, whether positive or negative.

[Anonymous No.6642480]

>>6641121
Kill yourself jew. Art is and will always be defined by being aesthetically pleasing and soulful, Please jump off a building. The word art got perverted over the years.

[Anonymous No.6642486]

>>6639298
Replace AI with photoshop and see how dumb this whole post looks

[Anonymous No.6642499]

>>6642480
>Art is and will always be defined by being aesthetically pleasing and soulful
by who?

[Anonymous No.6642501]

>>6642499
By anon no. 6641121 of course

[Anonymous No.6642512]

>>6641121
my grandma's funeral is art

[Anonymous No.6642538]

>>6642499
Literally all of humanity until a few decades ago nigger. There is a reason nazis called it degenerated art what the jews called "abstract".

[Anonymous No.6642554]

>>6642538
Nazis lost the war, their rules and thoughts have no weight. Try going to an alternate timeline

[Anonymous No.6642749]

>>6640621

[Anonymous No.6642755]

>>6642538
>a bunch of inbred hun rape babies = all of humanity
That's some serious keks there

[Anonymous No.6642758]

>>6641121
wrong, a sunset or a forest is not art, there are many things in day to day life that we have emotional responses to that is not art

[Anonymous No.6642762]

>>6642758
>A sunset or a forest is not art
According to /ic/, this is not art. Until you add an anime girl :^)

[Anonymous No.6642784]

>>6640678

Nah... There's nothing new in the last 3 years other than controlnet. Everything else is just the quality of data stolen thats finetuned into the models. The data is everything. The tech itself doesn't improve much. If there's even 1% of improvement techbros will hype it up. Take away the stolen shit and you'll end up back to square one. The reason why people who put money in it will do whatever they can until they get their money worth. Adobe AI is the only one that's trying to go a different direction with their smaller dataset. They're not the good ones but at least it's heading towards being an actual tool artist can use. StableDiffusion on the other hand is just a tool for stealing.

[Anonymous No.6642848]

>>6639326

[Anonymous No.6642852]

>>6640862
I guess but rendering at that level is a bitch

[Anonymous No.6642928]

>>6639298
Ok I'll bite. Here's how AI works in detail.

[Anonymous No.6642929]

>>6642928
The underlying technique the AI is designed around is called "Stable Diffusion", which is basically a really fancy resolution-increasing/unblurring algorithm. It was originally designed to make blurry photographs sharper. It takes a blurry image, then guesses what the blurry parts are supposed to be, and fills in the blurry parts with sharper details based on its guess. Turns out, if you use it on images that are _completely_ blurry, it'll just make up an entirely new image from random guesses, so that's what all the AI image generators are based on.

For the AI to be able to guess what a blurry group of pixels are supposed to be, it needs to be taught that a particular set of blurry color strokes, in a particular part of the image, are supposed to be a specific thing. This is done by training the AI on a bunch of example images, having the AI blur them a lot, and then comparing the before and after to see what the blur's supposed to look like when it's sharpened. For example, a skin-colored splotch near the top middle of the image is usually a face, so by feeding the AI a bunch of example images of people, the AI will see a lot of skin-colored splotches near the top that came from sharp pixels that look like faces, and then whenever it generates an image later, it'll try to draw a face there if there's any skin-colored splotches around that area.

[Anonymous No.6642931]

>>6642929
This would let the AI generate random images from random blurry colors, but it's more useful if you have some control over what it generates, so that's where 'tagging' or 'prompts' come in to play. When the AI is trained on images, the person training the AI also provides a description of what's in the training image, like "a woman sitting on a chair" or "a man in a spacesuit standing on the moon". For newer models this is done with "tags," like "woman, sitting, on chair" or "man, spacesuit, standing, moon". When the AI learns what those images look like blurred, it also stores that knowledge in a way where it'll only use that knowledge if the same words are used. So if half the training images are of women, and those training image was described as "a woman ___", it will only generate a woman if the user asks the AI for "a woman" when generating an image. This description you request from the AI is a "prompt."

The part of the AI that gets trained with all these images is called a "model," or "checkpoint" (since when you share the AI's file it's like a snapshot of all the training up to that point). Training a model to the point where you can get detailed images that accurately match prompts takes a long time, a lot of computing power, and a lot of training images and descriptions to feed to the AI, so the most popular models tend to be made by groups with a lot of resources and then shared around. Models like the original "Stable Diffusion", "NAI", "Protogen," and others serve as the base model for most image generations. Once these models were set up, though, it was much less work to train them just a little bit more to learn specific art styles, characters, or concepts. Many derivative models exist that take one of these base models and then train it further.

[Anonymous No.6642933]

>>6642931
These models can be "mixed" or merged together, as long as they come from similar baseline models. These mix models have become fairly popular in their own right for being more flexible than other individual models. "AbyssOrangeMix,", "Counterfeit," and "Deliberate" are commonly used merged models.

In addition to full models/checkpoints, there are other ways of doing smaller sets of AI training and sharing that in a reusable way. The first major method was called "textual inversion," "DreamBooth," or "embeddings," which worked by copying the baseline model, training it on a new concept, and then making that new model only respond to a particular prompt keyword. It was just as powerful as a standalone model, but gave you the flexibility of turning it on or off depending on whether you included the keyword in your prompt. These textual inversion models tended to use a lot of memory and storage space, though, so they've become less popular over time.

The next major technique pioneered by the Novel AI team was "hypernetworks," which were mini-models designed to "plug in" to a base model and tweak the results it generated instead of generating the image entirely by themselves. These were less accurate than textual inversion models since they only affected a part of the base model's generation process instead of the whole thing, but they were significantly smaller in size, and you could apply multiple of them at once as a result, creating new ways of mixing and matching models.

[Anonymous No.6642934]

>>6642933
The idea of smaller "plug in" models was further refined with "LoRA" models, which were more powerful because they could affect more places in the image generation process, including teaching new prompt tags, but kept the relatively small file size. These were small, powerful, and easy to share, so people quickly produced many character and style LoRA models teaching the AI specific things that then could be combined together.

LoRA models have been further improved with "LoHA"/"Locon"/"Lycoris" models, which are an improved version of LoRA models that affect even more of the base model's generation process and can be trained to affect the overall style of the image less than a LoRA or hypernetwork, making them more reusable.

In addition to text-based prompts using "txt2img," the stable diffusion technique is capable of also using an image to guide the result, so-called "img2img." This works by blurring the provided image a certain amount, and then unblurring it again using the algorithm and guided by a text prompt. This method can also be used for "inpainting," where only a section of the input image is affected and the rest is left alone, and "outpainting," where the AI tries to generate more stuff outside of the bounds of the input image.

[Anonymous No.6642973]

>>6639306
>Well now the gates are open and any one can create and express whats in their mind and it wont be as profitable for me clout/attention wise, I'm now a nihilist
lmfao sad

[Anonymous No.6642977]

imagine shidding and farding ur pants that people can now just type a few words and click a few times to produce better art than the stupid postmodern hacks who splatter paint on a canvas to have some jew throw tons of money at them to subvert everyones concept of beauty

>>6640692
if you ask chatgpt about certain obscure medical things for example, you will see that it genuinely seems to understand what its talking about and can give people a deeper level of knowledge and deeper insights than any single human on the face of the earth, by miles

[Debater No.6642993]

>>6642929
>>6642931
>>6642933
>>6642934
I could ask why type all this shit out when you basically saying it more tag specific with a zoom in function. But not you retards to consolidate your fucking words

[Anonymous No.6643064]

>>6639326
> you can't even generate the most simple stuff like a girl spilling a cup of coffee after bumping into someone
I mean
A: yeah you can within like 10 minutes of dialing in a prompt
B: It’s improving exponentially
C: the rest of your post is cope about potential gubment regulation

[Anonymous No.6643067]

>>6643064

?? HUH

[Anonymous No.6643070]

>>6643067
What’s confusing? It’s not difficult to generate images of a qt girl bumping into someone and spilling coffee on herself.

And this post it’s not even hard to make gifs of Emma stone getting served coffee by an adorable coffee demon.

[Anonymous No.6643072]

>>6643064
This is barely qualifying for what that anon asked

[Anonymous No.6643074]

>>6643070
*and at this point

[Anonymous No.6643079]

>>6643072
It’s a second take of entering stuff into the prompt are you autistic?

It took 30 seconds to make something that would take take hours otherwise. Oh no it’s not perfect? Guess I got fucking hours of time to generate thousands of images to dial it in

[Anonymous No.6643123]

>>6643079
an artist can do it in less time and better. this is sad and the vomit you've posted is hilarious to look at in a bad way

[Anonymous No.6643185]

>>6643079
HAHAHAHAHAHAHA

AHAHAHAAAAAAAAAHHHHH thanks for the laugh anon, please try it with anime style next I need some good laughs

[Anonymous No.6643298]

>>6643079

generate another 3000 you might get one.

[Anonymous No.6643307]

>>6643079
/ic/ is known to crab anyone posting good work, you're on the right path to make it

[Debater No.6643316]

>>6643307
/ic/ crabs on everyone why do you niggas want to expect something else coming here spamming shit

[Anonymous No.6643335]

>>6643079

share your model I want to try

[Anonymous No.6643440]

>>6643064

[Anonymous No.6643488]

>>6643307
>"work"
>"make it"
I cannot actually believe I'm seeing some faggot hugboxing another faggot gacharolling pictures in the hopes of getting something good and calling that "work"
El em friggin' ayyo

[Anonymous No.6643558]

>>6643074
>>6643070
holy kino

[Anonymous No.6643570]

>>6642977
post modern """artists""" don't give a shit about you and your AI you fucking retard. The ones you leech off despise you

[Anonymous No.6643575]

>>6642977
>if you ask chatgpt about certain obscure medical things for example, you will see that it genuinely seems to understand what its talking about and can give people a deeper level of knowledge and deeper insights than any single human on the face of the earth, by miles
nice platitudes, but you sound like a colossal rerard. Explain in detail what are those obscure medical things, and how does it guve a "deeper level of knowledge". You think your overngineered chat bot is some fucking lovecraftian otherworldly being? I've noticed your shit pic is AI without even opening it

[Anonymous No.6643585]

>>6643575
Just cuz you're a fucking idiot doesn't mean everyone else is. https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45732.pdf

[Anonymous No.6643587]

>>6643079
Why is there Russian on her shirt

[Anonymous No.6643594]

>>6643587
she's fast

[Anonymous No.6643602]

>>6643123
You’re not an artist lol
Post your best art that you’ve produced on under 2 hours.

Also what’s your argument other than hopeless cope? None of this tech was even a thing a year ago and now it’s easy and consumer facing. And again: IT. IS. EXPONENTIALLY. IMPROVING.

You’re basically looking at early motion pictures and saying that movies will never be a thing because they don’t look perfect right now. As if time exists in a vacuum and the future doesn’t exist. As if something would happen that would make things like inertia and entropy stop existing.

[Anonymous No.6643617]

>>6643602
Eventually, we will all be dead and the sun will explode. Why get up in the morning?
It's funny how much it triggers you cretins that some people prefer drawing over prompting goyslop. Get a life, if you're actually even a human being.

[Anonymous No.6643619]

>>6643335
Idk I’m using mage.space which uses stability diffusion and the model version is one of the animu ones. I’ve found that for most porn and action type of prompts, if you do one of the anime model versions first you get better txt2img results, and then you can use a realism model version on the generated anime image to keep the framing and blocking and action but make it look realer

[Anonymous No.6643622]

>>6643617
> It's funny how much it triggers you cretins that some people prefer drawing over prompting goyslop
A: you’re an ESL foreign invader and you should go back.
B: Nobody gives a shit about how much you asiatic faggots pretend to draw. Being fascinated by AI isn’t an attack on you or your shitty /beg/ drawings you don’t even post, ya ngmi bozo larp

[Anonymous No.6643671]

https://youtu.be/NCuBalItZA8

It’s too late to fight it. Might as well be on roko’s good side at singularity

[Anonymous No.6643693]

>>6640621
The left cant meme

[Anonymous No.6643747]

[Anonymous No.6643750]

[Anonymous No.6643758]

[Anonymous No.6643767]

[Anonymous No.6643778]

[Anonymous No.6643803]

AI art looks kind of fun to be honest.

[Anonymous No.6643811]

>>6643488
>gacharolling pictures in the hopes of getting something good
I've been generating a bunch of ai stuff in the past week and this perfectly describes how it feels lol. You often can get something "good" but if it has to be specific you are fucked. That's why it's so absurd that these people say the ai makes what they imagined all along

[Anonymous No.6643814]

>>6643811

[Anonymous No.6643816]

>>6643811
skill issue

[Anonymous No.6643852]

>>6643622
Okay, Trigglypuff.

[Anonymous No.6643855]

>>6643440
This guy fucks

[Anonymous No.6643867]

>>6643747
>>6643750
What’s that using?
Mage.space says it has an img2gif option coming soon but I’m regarded on all the free pc shit rat meow.

When do you think will be the first financially successful feature leangth ai movie?

[Anonymous No.6643879]

>>6643594
Why I aughta

[Anonymous No.6643892]

>>6640438
>How'd the amateur beat it? By playing so amateurishly the AI didn't know what to do.
Makes sense. Even with humans, skilled competitiors are used to playing against other skilled competitors and tend to predict and account for the logical moves of other skilled players, so they can get thrown off by a newbie who has no idea what the fuck they're doing and often do wrong, completely unexpected things.

[Anonymous No.6644016]

>>6643585
you most definitely are a fucking idiot because you haven't answered me the question.

[Anonymous No.6644055]

>>6643602
Ok, consoomer.

[Anonymous No.6644079]

>>6640678
you realize it was shitty for years until they decided to use copyrighted works right?

[Anonymous No.6644089]

>>6643619
>>6643079
>>6643070
fucking kek
artists had nothing to fear this whole time.

[Anonymous No.6644100]

>>6644089
>artists had nothing to fear this whole time.
>He says that without being aware of the new models being constantly released
I was fortunate enough to gather some fully rendered illustrations being created right before my eyes with Miku scenes and themes at the fingertips, I'll spare you from the uncomfortable truth and leave you lulled under a blissful state of that lie of yours.

[Anonymous No.6644117]

>>6644100
>JUST TWO MORE WEEKS
>TRUST THE PLAN

[Anonymous No.6644140]

>J-JUST WAIT!
I've been waiting a year now and AI is still an incoherent mess. It just hides its mistakes better.
The models now are prettier, but they still don't understand anything they're making

[Anonymous No.6644143]

>>6644140
photoshop doesnt understand itself either

[Anonymous No.6644147]

>>6644117
>>6644140
You two need to accept the fact that not even a year ago the Ai had drastically improved in the span of a couple of months, and it will continue to add more to its library which will enable to create even more stunning illustrations from our wildest dreams. Say what you want about the slight mutations, or the structures/buildings not being fully in line with the vanishing point, or a distorted form not fully symmetrical to the figure it's the concept and idea that will appeal to people like me who yearn for more content of a particular character or a concept. With that said, have a piece of what was being produced a night ago this illustration is just a fraction of what the Ai is capable of my heart bursts with so much joy knowing this image is an expression from a kindred spirit who also wishes to see Miku in a world of our own.

[Anonymous No.6644153]

>>6644147
>Say what you want about the slight mutations, or the structures/buildings not being fully in line with the vanishing point, or a distorted form not fully symmetrical to the figure it's the concept and idea that will appeal to people like me who yearn for more content of a particular character or a concept.
The 'idea' you want to see more of is 'overrendered Miku with a nonsensical outflit/hair looks at a nonsense city'?

[Anonymous No.6644168]

>>6644153
>overrendered nonsensical features
>nothing makes sense
It's the immersion brought to you from a visual medium behind a gesture or expression, in other words it's what makes aesthetics and the art that communicates to you in another form that leaves you with an impression. I can't believe that I have to explain this to somebody that presumably creates art, if you still fail to understand what I am trying to convey here and continue to foam at the mouth at the Ai then I feel nothing but pity for you.

[Anonymous No.6644418]

>>6644168
We all have 'ideas', retard, that doesn't make you special. The point of art is communication of those ideas. If you're only fixated on the ideas, you don't give a shit about art, you just want to mindlessly consoom as many images as you want. My appreciation for art goes beyond that. I actually care about individual technical skill as well.

[Anonymous No.6644421]

>>6644418
>I actually care about individual technical skill as well.
I call bullshit, I've seen you fags do nothing but step over each other's dicks for the past ten years.

[Anonymous No.6644431]

I don't understand why people here seethe so hard over AI art and see it in a tribalist way where anyone interested in AI art must be obsessed with it and hate artists instead of just someone being a well rounded individual with many interests being intrigued by new technology. If you make art because you enjoy the craft and want to share YOUR ideas how is AI hurting you? We had chess AI for a while that could beat the best chess player and yet chess is still huge. People will always value art created by other people. Artists should also see AI as another tool in art creation. They could let it inspire them and then use their skill to clean up the output.

[Anonymous No.6644448]

>>6644431

scraping data without consent.

not allowing artist to OptIn instead make artist optout their work manually.

releasing it to the public allowing thief, scammers and any type of scumbags to do damage and profit off other artist work.

not a tool. good artist doesn't need this and bad artist will never improve if they use this.

>They could let it inspire them and then use their skill to clean up the output.

like a factory worker.

you don't understand art and probably never will. not because you are incapable but because you won't do it. don't try to tell artist what to do with this tech when you yourself don't understand artist and they way we create work of art.

[Anonymous No.6644475]

>>6644448
>scraping data without consent.
you don't need consent to scrape or use freely accessible data
>not allowing artist to OptIn instead make artist optout their work manually.
opt in to what? being inspiration for art? get real.
>releasing it to the public allowing thief, scammers and any type of scumbags to do damage and profit off other artist work.
what are you talking about? you can say the same thing about releasing pencils to the public
>not a tool. good artist doesn't need this and bad artist will never improve if they use this.
it objectively is a good tool and there are now millions of new artists thanks to AI, and most of them are way better than you(feel free to prove me wrong, post your work)

[Anonymous No.6644492]

>>6644475
>you don't need consent to (...) use freely accessible data
Wrong. Scrapping is only ok because it's pretty much just accessing it.

[Anonymous No.6644600]

>>6644492
What do you mean wrong, you're saying the same thing. If the data was behind a paywall or even a simple login there could be some argument but it's literally just there, freely available on the public internet.

[Putin Pedofil No.6644609]

>>6644055
Nobody reading all that shit.

[Putin Pedofil No.6644612]

>>6644140
You prolly just dogshit at prompting. You doing shit locally or did you like type some nonsense into a free dall e thing or summink?

[Anonymous No.6644740]

>>6644475

>you don't need consent to scrape or use freely accessible data

if your house is unlock I'm free to come in and take your shit? or are you going full retard by saying shit like "AI CAN SEE JUST LIKE HUMAN"??

>opt in to what? being inspiration for art? get real.

"consent" you want to use my work? ask me if I want to. easy. something every normal parents in the world taught their child.

>what are you talking about? you can say the same thing about releasing pencils to the public

obviously you get it but you don't want to admit it. don't brush it off. you know the fuck up shit people do with it.

>it objectively is a good to--

not a tool. ai user are not artist. you are not an artist. your image is made by an app. your prompts belongs to you. but not the end result. Show me your full process and I'll tell you which part is yours.

[Putin Pedofil No.6644763]

>>6644740
>Reddit spacing
I’m normally not the type of feller to make fun of Reddit fags but you’re both a Reddit invader and an ESL invader, and you’re posting anonymously like a punk bitch that can’t draw for shit. If I were a gamblin man I’d bet you don’t even give a wet squirty shit about art or art production and you’re only into this for the culture war.

[Anonymous No.6644820]

AI is here to stay, the sooner you accept it the better outcome for you as an artist it will be.

Is the art detailed and specific, ehh so so
You can tweak the options and use different checkpoints with ControlNet but the result is always distinguishable from a real piece.

Porn junkies are having the most benefit from it since they don't really care much about the finer details, as long as the general feel of it is on point.

In my own opinion going against the AI move is like saying that the invention of cars will take carriages away, it's true but it's pointless to rebel against it.
Learn about it, find creative ways to incorporate it into your workflow so you can express yourself more without suffering from general artistic health problems.
Speed up at least the creative idea generating process.
We already had a win in the courts where they said the AI art can't be copyrighted.

As an example, I'm personally using it to make stock photos for web concepts and when I have a boring client for design I create prompted low fidelity artwork to inspire myself.
Is it ground breaking, nope but it saves me up 2 + hours of being an indecisive prick and I can flesh out my clients ideas faster since a lot of them aren't really good at describing the ideas in their heads and often they don't really know what they want.

[Anonymous No.6644867]

AI art is a mirror. It exposes how shallow and boring human art is.

[Debater No.6644871]

>>6644867
If that's how you feel looking in a mirror than sucks to be you than

[Putin Pedofil No.6644922]

>>6644867
>literal projection
You addicted to staring at a black mirror br. Quit putting your autism on all of humanity

[Anonymous No.6644967]

>>6644100
>over rendered goyslop will now be slightly more over rendered

[Anonymous No.6645040]

>>6644448
dont fall for the
>"like UGHH I just DoNt UndErSTaND WhY ArtIstS ArE sO Up in ARms"
bait. Its a common line the shills use

[Anonymous No.6645063]

>>6644600
You third worlders just cant understand basic concepts like licensing or ownership, can you

[Anonymous No.6645068]

>>6645040
kek
same with anyone using muh workflow meme.

[Anonymous No.6645110]

>>6644600
>"freely available on the public internet."
>falsely equating the public using the internet to browse copyrighted works as being the same thing as copyrighted works that are considered public domain
Its statements like these lets me know that artists had nothing to worry about. To think that these pajeets actually thought they said something and even felt legally justified in doing what they're currently doing puts me at great ease

[Anonymous No.6645149]

>>6644475
>you don't need consent to scrape or use freely accessible data
the public at large was forced to make their data 'freely accessible' because it's then sold to advertisers. 'just don't use the internet' is a retarded fucking argument and you deserve to get your skull caved in if you use it unironically

[Anonymous No.6645289]

>>6644740
>if your house is
Stopped reading here.

>>6644867
It exposes the complete opposite for me.

>>6645063
>you wouldn't download a JPEG
oy vey

[Anonymous No.6645306]

>>6645149
the fuck are you talking about

[Anonymous No.6645413]

>>6645289
>oy vey
Let me guess, you want to "abolish copyright" too?

[Anonymous No.6645417]

>>6645413
It won't be relevant either way. AI will let anyone almost instantly create whatever they want eventually.

[Anonymous No.6645451]

Jesus how many cope threads will there be? Technology has already proven to overtake manual labour over and over again, so it will inevitably happem with drawing again, so shut the fuck up and enjoy drawing for the sake of drawing and stop crying like a little bitch

[Anonymous No.6645455]

>>6645417
And none of it will have any value whatsoever. Not just monetary value.
It's not yours, none of it is. Any idea executed through AI loses any meaning or good it might have had otherwise.
That's what you ideafags doggedly refuse to understand, and why you'll never succeed or find meaning in what you do as long as you don't invest yourself into it.
You should let go of this vile consoomer mindset. Try creating something by yourself and you will understand. And I mean by yourself, not through a program peddled by the same 3 big tech companies that does 99% of the work.

[Anonymous No.6645459]

>>6645455
As it should be, I'm really glad it was ruled that AI generated imagery has no copyright as that allows anyone to take anyones style/drawings and strip away any copyright from them by simply running it through img2img, and then use them for whatever you want

[Anonymous No.6645462]

>>6645459
not everyone aspires to be a subhuman parasite like you, good luck with that mentality though

[Anonymous No.6645463]

>>6645462
subhuman parasite? or simply one that knows how to play the game, and exploit retards like you

[Anonymous No.6645469]

>>6645455
>And none of it will have any value whatsoever. Not just monetary value.
To you.
>It's not yours, none of it is. Any idea executed through AI loses any meaning or good it might have had otherwise.
>That's what you ideafags doggedly refuse to understand, and why you'll never succeed or find meaning in what you do as long as you don't invest yourself into it.
Nonsensical opinion.
>You should let go of this vile consoomer mindset.
How ironic.
Try creating something by yourself and you will understand. And I mean by yourself, not through a program peddled by the same 3 big tech companies that does 99% of the work.
Post your work and show me what you're talking about.

[Anonymous No.6645593]

>>6645459
>strip away any copyright from them
get a load of this fucking retard

[Anonymous No.6645598]

>>6645593
that's what it means, as the image was entirely produced by computer, it was created by noone, thus there's no copyright to it, no matter what the training data was, that's what the precedent set with the case of the comic

[Anonymous No.6645601]

>>6645593
who do you sue if someone img2img'd your art? according to the law a robot did it.

[Anonymous No.6645630]

>>6645598
reread his post retard
>>6645601
you can't strip someones copyright lmfao. Thats not how that works you streetshitting baboon

[Anonymous No.6645633]

>>6645630
no one said you could. but there's no infringement since a robot did it.

[Anonymous No.6645634]

>>6645630
Not directly from the same image, but by producing a 99.9% derivative that was generated entirely by computer

[Anonymous No.6645659]

>>6645633
>no one said you could.
He literally just said it you fucking bum. Final (You)
>>6645634
> producing a 99.9% derivative
Then you're not stripping anything. In fact they could just formally register it with the copyright office and DMCA you to the point of getting you banned of any site you try to reupload it to. And since you have no proof of work and the AI work doesn't qualify for copyright, you'd be fucked.

[Anonymous No.6645663]

Just a reminder while AI art as a proof concept has been out for year or two Dalle2, Midjourney and Stable diffusion came out in June - August

It hasn't been even a year and the artist asshurt and seethe is going on... get over it.

[Anonymous No.6645665]

>>6645659
you're really missing what we're saying, idiot

[Anonymous No.6645666]

>>6640614
If you pretend that your AI prompted pajeet generation has deep intellectual meaning I guess.
90% of modern art is mainly about intellectual circlejerking to convince the minds of the art industry. Spiritual value of art has been lost, with some exceptions (like Bill Viola, first name coming to my mind)

[Anonymous No.6645672]

>>6645663
>ai art been out since 2015
>was complete shit until it started using copyrighted images
this will end well

[Anonymous No.6645770]

>>6644055

[Anonymous No.6645782]

>>6645663
no host is gonna get over your parasitic ass, ever, ever. stop trying to think that "this is just like photoshop yo, people will get over it". you are absolutely delusional if you think this
>get over it.
the AIcuck rhetoric of "it will only get better, it cannot be stopped" is just a code word for "submit now!". stop bitching that people are defending their work and pushing away the parasites that go after it, your commie mentality is never going to apply to humans. you are seemingly too inept to even understand why the humans who create are doing it in the first place

[Anonymous No.6645790]

>>6645782
It's exactly like photoshop and people will get over it. Photoshop uses AI.
GET OVER IT.

[Anonymous No.6646068]

>>6639523
>I have a friend that uses AI art and he said that he can't get all the poses he want.
Tell your friend he's a retard for still not using controlNet

[Anonymous No.6646085]

You will never be a real artist. You have no skill, you have no talent, you have no discipline. You are a consumer of algorithmic noise twisted into a crude mockery of a real draftsman's perfection.
All the “validation” your images get is two-faced and half-hearted. Behind your back people mock you. Your parents are disgusted and ashamed of you, your “friends” laugh at your ghoulish jpegs behind closed doors.

Artists are utterly repulsed by you. Thousands of years of practice have allowed artists to sniff out frauds with incredible efficiency. Even AI images that “pass” look uncanny and unnatural to an artist. The shiny, android-like skin is a dead giveaway. And even if you manage to convince a drunk guy that your image is real, he’ll turn tail and bolt the second you ask him to pay for your generic, effortless digital vomit.
You will never be happy. You wrench out a fake smile every single morning and tell yourself AI is going to get better, but deep inside you feel the depression creeping up like a weed, ready to crush you under the unbearable weight.
Eventually it’ll be too much to bear - you’ll buy a sketchbook, pick up a pencil, draw a loomis head, and plunge into the cold abyss. Your parents will find you, relieved that they no longer have to live with the unbearable shame and disappointment. They’ll frame your drawing signed with your birth name, and every passerby for the rest of eternity will know a real artist created it. Your body will decay and go back to the dust, and all that will remain of your legacy is your art that is unmistakably hand-made.
This is your fate. This is what you chose. There is no turning back.

[Anonymous No.6646095]

>>6646085
only bad programmers worry about gpt
only bad artists worry about stablediffusion
good craftsmen will adapt

[Anonymous No.6646125]

>>6646085
>fake artists are better than real artists
weird world we live in

[Anonymous No.6646128]

>>6645790
>swallow your roach milk and get over it!!

[Anonymous No.6646140]

>>6644055
Legendary post

[Anonymous No.6646152]

Just gonna stop right here to tie my shoe and... whoops looks like I dropped something, there it goes, someone catch it please!

[Anonymous No.6646158]

Wow jeez am I a klutz or what???

[Anonymous No.6646160]

Oh sheewawa eeesh I just can't seem to help myself today, I am SOOOO embarrassed >///<;

[Anonymous No.6646162]

OH NO! NOT THE SECRET DOCUMENTS!
THEY CAN NEVER KNOW ABOUT THE LATENT SPACE, SOMEONE GRAB THEM

[Anonymous No.6646171]

I think I managed to get this article cut and pasted to a size 4chan will accept, so here goes.

[Anonymous No.6646178]

So yeah, OP is entirely correct. ML models basically take a set of data and generate "fake" data, or interpolations, in between the original data points.
So these machine learning models took the original works of artists as well as countless photographs, ground up into a slurry and packed it full of fillers- Metaphorically speaking, of course. Calling it "goyslop" is actually far more accurate than you might have imagined to begin with, painfully so.

[Anonymous No.6646329]

>>6643064
LOL

[Anonymous No.6646344]

>>6646160
>barely any resemblance
Oh geez I guess I gotta go sue artists if their work looks anything like the reference!
>>6646162
>>6646171
>>6646178
Which by principle works similar to the brain you may not own retard? All your thinking is influenced and spliced with memories of things you have seen in your life, so you are quite literally the walking goyslop spitter, brainless one at that

[Anonymous No.6646346]

>>6646085
>wanting to be artist
No thanks I’d rather not starve, doesn’t sound fun

[Anonymous No.6646351]

>>6646085
the whole point of AI is to not be an artist, artists are inherently inefficient workers, and the whole point is to replace them as much as you can, so by aspiring to be an artist you are quite literally striving to be an inefficient, antiquated worker

[Anonymous No.6646352]

>>6646344
>barely any resemblance
You have problems bro
"results indicate that at the initial visual detection stage, facial processing is inefficient in schizophrenia. By isolating face detection from other aspects of face recognition, this study identifies a face-specific visual deficit in schizophrenia, which may ultimately contribute to impaired face-related cognitive and emotional processing and social interaction."

[Anonymous No.6646356]

>>6646352
>different style of facial hair, the side part is entirely different
>different angle of face
>different nose
>very different hairstyle
>different look direction
>facial bone structure is more narrow in cheek area
NOO ITS MY REDDIT HERO, LE HECKIN WHOLESOME 100 KEANU
I wish you morons paid as much attention to your own arguments as you pay to spotting any minor flaws with AI art lmfao

[Anonymous No.6646364]

>>6642973
DO NOT REDEEM SIR

[Anonymous No.6646366]

>>6646356
>Vague
>Eyes and eyebrows line up perfectly, the most recognizable aspect of a person.
Beard also a match in the soul patch area

[Anonymous No.6646395]

>>6646366
Not really though? The eyebrows are thinner in case of the photograph, and the angle is different, as you can see the eyebrows go closer towards eye as the image transitions from painting to photo
Also the beard lacks the connection between the main part, and the part under the lips, one that Keanu has

[Anonymous No.6646444]

>>6646351
Them explain why aifaggots try so hard to make their slop pass as real art

[Anonymous No.6646448]

>>6646395
I possibly could have scaled the images closer. The fact so much aligns though, the eye shape and position, the brow angle, nose width/length, the mustache coverage, soul patch width(sure, doesn't connect), the cheek(our right) shape, cranium size the facial thirds proportions... The irrefutable fact AI is known to do exactly this from examples like afghan girl and great depression lady. do the math.

[Anonymous No.6646558]

>>6646356
AIcucks cannot even seem to grasp what artists go through when they ask for criticism on their art. hell, AIcucks get offended at any form of criticism and call it nitpicking. this is why ywnbaa

[Anonymous No.6646576]

>>6646351
even digital artists don't talk about traditional like this. AIcucks are peak clownery

[Anonymous No.6646577]

>>6646351
>antiquated
>keep using them in your prompts
If they're antiquated, why do you need their art?

[Anonymous No.6646850]

>>6643079
>are you autistic?
He might not be, but those "people" in your "art" clearly are.

[Anonymous No.6646873]

>>6646558
They're mad because they don't have the skills to fix the flaws in the turds that the Ai made for them

[Anonymous No.6646917]

>>6646351
How are you going to replace me? Is a guy who can't do anything but type prompts going to take my job? Lol.

[Anonymous No.6646951]

>>6643892
Except that's not how the AI was beaten. Do you guys only hear what you want to hear!

[Anonymous No.6647214]

>>6646558
>>6646873
AI makes better art than you just get over it.

[Anonymous No.6647229]

>>6647214
There many people who made better and worser art out there that is somehow popular. Here's your daily reminder that this shit exists and been sponsored even

[Anonymous No.6647557]

>>6639298
>They cant understand concepts
Thats the opposite, current ML systems understand concepts really well, which is the reason they can compress so much Data. GPT4 knows almost anything in the internet to some extent, even if storing all of this data should be impossible.
>Next word predictor cant understand
The whole next word prediction is just a way for them to generate text and do the training. Every time it writes a new token, LLM sends the signal through its whole "brain" in order to predict the best next word. You could have actual human being locked in some box and tell him to read the whole text and then predict next 10 words, that text could be anything from book mockup to complex logical riddle with answer being the last 10 words. You could even have there 5 people communicating with each other about what the next 10 words should be and it would still end up as next word predictor, even if those people locked up there are real people with real thoughts, emotions, consciousness and thinking patterns, they are still collectively just next 10 word prediction club. This is the whole idea behind machine learning, you take systems impossibly complex for humans to code by hand, give them random weights and then let them train until the random weights connect together into system that can reason, or at least mimic it well enough to pass all the training data.

[Anonymous No.6647576]

>>6647214
Ai doesn't make art checkmate atheists

[Anonymous No.6647578]

>>6643602
>This thing didnt exist year ago, now it is consumer facing
Wrong dumbass. Dall-E 1 is more then 2 years old, in mid 2022 it just got released into the public and became easy
https://openai.com/research/dall-e
>Exponential growth
Midjourney was able to do photorealistic stuff in V-3 if you look back at their older images. Now they just improved the lighting to look less fake and more like photos because they stopped relying as much on artists and started using more photos as the training data. Yes, they fixed the hands, in most instances, but that took almost half a fucking year, just to fix that problem. That sequence of pictures of naked women looks just as shit as the SD attempts at making text2video by generating each frame by itself were. Your animation didnt improve in the slightest. At least use the more respectable text2video ones that blend the frames between eachother. They still look just as slimy as the ones from Meta half a year ago, but at least better then those flickering images you posted
>>6640678
>What did AI art look like 5 years ago?
Well, there was this one shitty AI generated painting sold in 2018 on and auction before the whole people realized that prompters need to do very little. But that was pretty terrible and not using any of the techniques that are being used now. Instead lets rather look at picture made 4 years ago from thispersondoesnotexist.com. Even back then it was able to generate perfect photorealistic faces. And even that was just higher definition upgrade to 2017 research that figured out how to combine characteristics from datasets to make combinations not found within the dataset. Expand this idea to every word in dictionary and you got Dall-E.
https://www.kdnuggets.com/2017/11/infogan-generative-adversarial-networks-part3.html

[Anonymous No.6647580]

>>6646576
You really think these are the people using Stable Diffusion and then posting it on DA? They are tourists who came here to insult artists.

[Anonymous No.6647583]

>>6640678
AI generation got faster and broader(almost all) subjects, but it's not exponential progress. This was done training on just Rembrandts work with AI facial recognition assist. This was also super high res, higher than 8k with full 3d height map brush strokes.

[Anonymous No.6647593]

>>6647578
> Wrong dumbass. Dall-E 1 is more then 2 years ol
Post dall-e txt-2-gif

[Anonymous No.6647634]

>>6647583
Holy shit that pic, Ai art got worse.
>soul vs soulless

[Anonymous No.6647683]

>>6646152
>hands fucked up again
>the more you look the more problems you see

[Anonymous No.6647684]

>>6647583
the left actually looks more like a human created art than the right.
this has to be a troll...

[Anonymous No.6647708]

>>6647583
The AI has a tendency to over contrast and over saturate. The old one on the left it's way better than the new one on the right

[Anonymous No.6647718]

>>6647684
>>6647708
the text on the image is meant to be mocking people who thought AI image gen just showed up in 2021 and has exponentially gotten better since then.
How do I make it a more obvious? or should I leave it as is?

[Anonymous No.6647726]

My reaction every time I see Ai art is "why would anyone want this"? None of it looks like something something would actually desire and consider acceptable to consume as an artwork. I wonder what percentage of AI art that gets posted has every been saved to someone's collection, it must be vanishingly small compared to real art right?

[Anonymous No.6647727]

>>6640614
>>6640621
NPC mind

[Anonymous No.6647728]

>>6647593
I first saw this in 2019. not dalle but same principle.
https://medium.com/syncedreview/deepmind-dvd-gan-impressive-step-toward-realistic-video-synthesis-12027d942e53

[Anonymous No.6647870]

>>6647593
>Post text-2-gif
Nobody had text-2-gif because it was not something people cared about. Also nobody has access to Dall-e 1. I can however show you closed beta text-2-video that is almost year old. Look at the latest text2video and compare them with results found in here https://makeavideo.studio/.. Aside from individual frames looking better because in the meantime there was breakthrough in generating images much faster thus allowing them to be used for a video, there is no difference if not signs of regression in terms of interpolation and video cohesion.

(File was too big, so I didnt bother sending the gif of boat from the Meta demo I linked, too much work for one post and making it smaller would make it look worse)

[Anonymous No.6647873]

>>6647583
This. The only reason text2image gotten any recognition was because it is in theory able to do anything, not just one thing you train it super hard on. We had photorealistic image generators multiple years ago, the thing that separates text2image from the rest is that it can produce anything you can ask it for like an artist, so it actually hurts upcoming artists, instead of just making pure mimicry of one person style that you train it on with tons of learning data required. It gave midwits or even just retardwits the availability, because not everyone is able to train whole model on just one person and then fine tune it to produce something resembling it.

[Anonymous No.6647890]

>>6647726
Outside of stock pictures.used o illustrate a book or some shitty article, AI dogshit only has a sliver of value for whoever generated it. It is spam.

[Anonymous No.6647896]

>>6647870
> Nobody had text-2-gif because it was not something people cared about
It absolutely is. Txt-2-img is just a stepping stone to txt-2-video. Why do people struggle with this?
It’s the beginning of holodecks ffs
> Also nobody has access to Dall-e 1.
Well why am I being fucking argued at then? Reminder that my original statement was “this technology wasn’t a thing a year ago and now it’s easy and consumer facing.”
Other than being autistic about the fact that some super-computer only accessible by some corporate research groups technically existed years ago, nobody is disproving what I fucking said.
> I can however show you closed beta text-2-video that is almost year old.
Which doesn’t disagree with anything I said
> there is no difference if not signs of regression in terms of interpolation and video cohesion.
Lol ok.

[Anonymous No.6647898]

>>6647890
I doubt real artists will lose out on anything but $10 commission from this. The people using ai now are people who would have used google images to get what they wanted before. Artists who sell their own art will never want to touch AI or else everything they ever make will be assumed to have been AI generated and their brand will be trashed. Only in commercial art will there be any affect, in the same places and capacities where photobashing was already rife.

[Anonymous No.6647900]

>>6647896
The watermarks are a nice touch on these totally not stolen images. Stock image sites are absolutely going to wreck ai companies in court.

[Anonymous No.6647918]

>>6647898
This is what I was thinking as well but the second you put that idea forward, AIggers come out of the woodwork to say you're coping.
Image gen tech is fucking useless to make anything but the most generic, derivative shit, or obvious bootleg. The only reason so many people drank the big tech AGI meme kool aid is because Silicon valley propaganda is on overdrive. After the likes of web3 and VR have failed so miserably and investors fucking off, they're all jumping on the AI grift. Its in their best interest to make an army of useful idiots peddle the idea that this new thing is somehow exponentially improving and will put everyone put of a job in 5 years while all they really got is an elaborate plagiarism machine. Cue all these tech startup roaches currently backpedalling hard under a growing pile of litigation as they realize that just maybe people might have caught on that using literal billions of licensed pictures in their data wood chipper without asking anyone (under the flimsy pretenses of fair use or for-research dataset) was a jewish scheme.

[Anonymous No.6647927]

>>6647896
>nobody is disproving
Other than the person who fucking disproved you, sure.
>Le researchers! Super computers! Technically...
It generated video on a single TPU. You can hire them for a few dollars(cloud).
The training is much more costly, speaking of super computers, stable diffusions training was done on 4000 GPUs. If they kept it closed beta(which I was in btw) would you still argue that ai image generation was not a thing yet? Your argument came in two parts, the former thoroughly disputed. "Easy and consumer facing"is fairly irrelevant regarding evaluation of tech progress. Can robots do parkour?
>No! unless you refer to closed source robots at high tech labs! So they can't!

So anyway, exponential it is not.

[Anonymous No.6647963]

>>6647898
I think more than anything it'll be used for weird fetish porn, stuff that's hard to get artists to draw.

[Anonymous No.6647977]

>>6647896
Here is the thing, your SD generating each frame on its own is going nowhere. Actual text2video that Meta and other companies use uses interpolation to remove the flickering. This text2gif where you generate frame by frame is going nowhere and looks terrible. I am not denying that text2video wont ever be a thing, I am just saying that this way of producing them is going nowhere.

Also, good text2video might be actually years away from now. Stop looking at text2img progress and just look at image generation progress as a whole. >>6647583 >>6647578 are examples of already high quality images being generated by AI many years from now, the whole "boom" in 2022 was just with AI being able to do it with all variety of images. Now show me high quality limited videos made by AI. Show me where AI is able to not flip legs when walking, or small strands of hair/grass not melting together.

[Anonymous No.6647985]

>>6647963
Weird fetish porn will be the last to go ironically. Most models will think Vore, Bodypart detachment, mutants and other weird stuff to be artefacts. The specially trained models on some of those output worse quality images because they go against its initial fine-tuning. Go look at guro or decapitation images on Pixiv and you will see that the images generally have more artefacts, especially around the decapitation part. And that is weird fetish with still enough training material, now do it for other fetishes. If there is not enough fap material of it around, there is not enough material to train it on period. Though you can get these weird fetishes by accident in the earlier models.

[Anonymous No.6647989]

>>6647684
No it is real https://news.microsoft.com/europe/features/next-rembrandt/.. It even was made on fucking paper with extruded paint perfectly mimicking trad art, not just fucking png file. Its almost as if they made this research and purposefully hid it away in order to then retroactively btfo tradfags who thought they would escape AI by using physical mediums.

[Anonymous No.6647998]

>>6647989
So more advanced printers

[Anonymous No.6647999]

>>6647963
This is the only reason I use AI. I will die before I allow an artist to know my name and permit them to render my deepest, darkest fantasies. They're mine and mine alone, but more importantly, I don't feel like drawing them myself.

[Anonymous No.6648002]

>>6647898
I need some of your copium anon, hopium has no effect on me anymore

[Anonymous No.6648028]

>>6647998
No, the image was actually generated by AI trained on all of his paintings and similar. In the post by Microsoft it says "The final piece is not a copy of Rembrandt’s work and it is not what he necessarily would have painted had he lived longer than he had – it is a visualization of data in a beautifully creative form. It is a powerful demonstration how data can be". The picture itself was indeed generated.

[Anonymous No.6648056]

>>6647999
Of course you dont. You're a /g/ tourist who cannot draw like every other AI faggot.

[Anonymous No.6648095]

>>6642762
nice thumb lmao

[Anonymous No.6648110]

>>6639298
it can do a lot of perspective correct shit so it somehow contains some form of understanding of 3D space

[Anonymous No.6648119]

>>6648095
This is your future.
>>>/g/93340129

[Anonymous No.6648182]

>>6647927
> So anyway, exponential it is not.
Tis. You’re not disproving a fucking thing. All you’re doing is being bad at comprehending time and scale and stuff. And most likely trying to cope with a reality where you put all your stat points into making the most easily-AI-able art forms

[Anonymous No.6648928]

>>6648182
see you in six months when AI still can't comprehend hands in novel poses

[Anonymous No.6648949]

>>6648182
>All you’re doing is being bad at comprehending time and scale and stuff
yet you are the one claiming>>6647583 was not a thing while it was a thing. stop drinking to koolaid. You are being marketed to by tech companies trying to raise hundreds of billions. Yes, It's getting better, incrementally. is there a sharp curve upwards on the way? who knows? there are large barriers that the best minds don't know a way around.

[Anonymous No.6649009]

So true sis!! Real artists have no reason to trace

[Anonymous No.6649010]

True artists are known to never trace

[Anonymous No.6649012]

[Anonymous No.6649015]

[Anonymous No.6649018]

>>6649015
>>6649012
>>6649010
>>6649009
i agree
none of these are true artists
thanks for proving our point faggot.

[Anonymous No.6649022]

>>6649018
Are you really shifting the goalposts?

[Anonymous No.6649025]

>>6649018
So none of these are real artists?

[Anonymous No.6649027]

[Anonymous No.6649029]

[Anonymous No.6649030]

[Anonymous No.6649033]

[Anonymous No.6649036]

[Anonymous No.6649045]

>>6649022
what goalposts? if you trace you arent a true artist
>>6649025
no
>>6649036
you cant prove that was "traced"
tracing and copying 1 to 1 is different.
if you have an image you want to copy and you sit it next to your page and you look at the picture and copy every single line that is not tracing, that is copying.
if you lay your page over the top of an existing image and draw on your page where all the lines are on the picture, that is tracing.
im not saying copying makes you a real artist either just that its not the same as tracing and that you cant prove that that gif was traced.

[Anonymous No.6649056]

>>6649045
There you go, shifting the goalposts again

[Anonymous No.6649064]

>>6649056
>shifting the goalposts again
prove it.

[Anonymous No.6649069]

>>6639303
>prompt: reply to OP, mock fp, insuiate tp is a homo

[Anonymous No.6649550]

>>6639298
yeah go back to drawing until AGI comes out in next 5 years, current AI is dumb but they can still improve the speed of developing AGI

[Anonymous No.6649556]

>>6640469
>6 days ago
hope you're still alive

[Anonymous No.6649644]

>AItrannies so mindbroken they lash out at tracers that everyone here has been hating on for at least a decade
clueless

[Anonymous No.6649944]

Man looks like that fanbox ban really anal raped the pajeets. Still obsessively coming in with the same eternally debunked arguments. Don't you guys have some proooompting to do? Crying here won't do shit.

[Anonymous No.6651042]

>>6649036
This is not traced, it just a homage.

[Anonymous No.6653587]

>>6647583
>with full 3d height map brush strokes.
???
And the left looks better. Saturation and contrast does not make shit look better.

[Anonymous No.6653611]

>>6649009
Nothing from the based was used for the final piece
A post isn't copyrightable. But composition based on someone else's work is. That's why he recreated his own composition then actually drew in and fixed the underlining issues the base had.

Aitroonjeetniggers all deserve universal erasure.

[Anonymous No.6653627]

>>6653587
>Saturation and contrast does not make shit look better.
it can

[Anonymous No.6653643]

>>6653627
Zoomer responce.

[Anonymous No.6653657]

>>6653587
yes. the image is made too mock aifags. I thought the '7 short years' bit was enough of clue...

[Anonymous No.6653941]

You can use every pose, every reference and get an 1:1 copy but with a different style etc, AI is amazing

pic related was made from a real model

[Anonymous No.6653967]

>>6653941
C'est magnifique, mais ce n'est pas l'art

[Anonymous No.6654152]

>>6649012
Sometimes I wonder why anybody would spend so much time and effort to produce such a generic piece of "art". And pixiv/deviantart/instagram/twitter is filled to the brim with this. With the advent of AI it got even worse, but at least I don't have to imagine someone sitting several hours in front of a screen tinkering with lines, colors and shading to get a dime-a-dozen anime pic.

[Anonymous No.6654177]

>>6653941
Girly, Truthfuly, I wouldnt clock your AI art as AI at all, yours looks like cis-Art to me

[Anonymous No.6656483]

>>6639326
>Also it can never create
Never say never to a technology

[Anonymous No.6656599]

>>6653941
cool can you generate her giving birth?

[Anonymous No.6656772]

>>6646351
>You will be made efficient, I just need to harvest you

[Anonymous No.6656774]

>>6646351
>You will be made efficient, I just need to harvest you

[Anonymous No.6656779]

The revolting gens always reek of falseflag bait, iToddler filenames don't help.
Digital artists talking about SD sound a lot like analog artists talking about digital, I just want a some wallpapers to put on shuffle and find the conflict from both sides hilarious

[Anonymous No.6656872]

>>6639298
>That's how machine learning and deep learning works, they find patterns in their dataset and try to create something "natural" without overfitting too much aka tracing 1:1
And how is this any different to how artists work? How often do you want to play that game and clapped? Also good luck fighting this losing war we literally have big corpos and literally every person who isnt in the "creative" field on our side.

[Anonymous No.6656899]

>>6656872
>And how is this any different to how artists work?
how many times will you ask this? it has been answered time and time again.
>we literally have big corpos and literally every person who isnt in the "creative" field on our side.
the delusion...

[Anonymous No.6656927]

>>6656899
>the delusion...
How am i wrong? Ask any normalfag and at best (for you) they wont care about this issue.

[Anonymous No.6656973]

>>6645790
Ai is the same as photoshop in the same way I am your father because I am male.
Literally!!

[Anonymous No.6656996]

>>6656927
I don't think your cell mates are a good sample.

[Anonymous No.6657051]

>>6656872
>AI
Interpolates bewteen pictures it was trained on. The easiest way to understand this is to make an image gen model from a single picture: it'll only be able to output noise resembling that exact same picture. Make it with two pictures, it'll only be able to generate a smear closely resembling either of the two.
>human brain
Creates an understanding of things present in pictures and uses that understanding to construct new pictures. We make use of structured mental concepts. When we see the picture of a sword we do not make a mental photography of that picture (like AI does), we understand its weight, how light interacts with it, how its forms give off the illusion of depth, etc.. Context. That means that we do not learn from the picture itself (like AI does), but the concepts present within that picture.

Before you even get started with this- no, AI models do not save pictures, they encore data weights closely resembling that picture in its gay virtual graph. This does not change anything previously pointed out.

You could boil it down to this: AI computes objective data, from a finite dataset comprised of defined, copyrightable pictures. We interpret elements from pictures subjectively.
These simple points that I just spent way too long explaining to your frogposting ass indicate that machine learning and human learning are obviously not analogous and pretending otherwise is playing dumb to push your dumb tranny agenda that benefits none but the same few silicon valley tech giants.

[Anonymous No.6657067]

>>6656996
Anon im on your side but you are not helping with this cope. Normalfags simply dont care or are in favor of it as long as it benefits them.

[Anonymous No.6657072]

>>6657051
>is playing dumb to push your dumb tranny agenda that benefits none but the same few silicon valley tech giants
>he doesnt know

[sage No.6657093]

'nother ai thread, 'nother doodle
drawing is fun.

[Anonymous No.6657097]

>>6657072
kek
Seriously though the only ones profiting from this are 4 californian jews getting away with laundering everyones data if they successfuly manage to divert everyone attention from actual issues with their sci-fi strawman bullshitting or fake idealistic memes
>this glorified autocorrect tool is just... like us... its basically sentient, look at how chatGPT tricked some user into doing things...
>AGI will end humanity!!! please stop us!! *pours another 500 mils into own AI company*
>muh democratization of le art (i will take a 10$ cut on this model you're using goy, also (YOU)'re the one liable if you plagiarize too much)
/g/edditors slurp up that shit like kool aid without a second thought. These people never stop for a second to think about why big tech even made this shit open source or why investors with a versed interest in AI development call for people to pretty please please stop developing AI because it's just too dangerous.
>>6657093
based catspammer

[Anonymous No.6657109]

My boy asmon is pro ai art or rather ai general and he usually is very reasonable when it comes to such topics. So i am as well, simple as that.

[Anonymous No.6657118]

>>6640614
>Top: money laundering scam "art"
>Bottom: money laundering scam "art"
What's the difference?

[Anonymous No.6657125]

>>6657109
based asmon lover
time to send him another GRIMDARK FRANCHISE X BALENCIAGA yt compilation, his 50mins reaction is gonna be epic

[Anonymous No.6658156]

>>6657109
asmon also almost puked from a tomato, he has the most retarded opinions on stuff and is a literal sociopath/psychopath that has no empathy towards others or animals, he even said it himself. Also he killed his mother because of neglect.

[Anonymous No.6658172]

>>6657109
he's pro ai only because it generates women with big booba, he couldn't care less about artists

[Anonymous No.6658234]

>>6657109
imagine worshipping this bucktoothed retard looking idiot and his dogshit opinions about anything.
guy's a complete sheltered fucking retard who changes his positions to the one that suits him the best, and then flips a shit to smallest amount of pushback...
Remember when this fucking loser had a panic attack because because of dmca during some special event bullshit.
Guy's a selfish dum dum with no real life experience outside of clickbait WoW non sense.

[Anonymous No.6658237]

this is the AI thread? first time I've seen a professional game use AI art unironically.(that rickmorty shoot game has some)

Hawken Reborn. It's so bad its funny. This game sucks balls btw, it's like they put the least amount of effort possible into every aspect of the game.

[Anonymous No.6658242]

>>6657109

The day I'll take any twitch streamer opinion seriously is the day the earth spin in reverse.

[Anonymous No.6658244]

>>6653941
wow so a beautiful babahiba hat

[Anonymous No.6658245]

>>6647557
It does not 'understand' anything

[Anonymous No.6658273]

.

[Anonymous No.6658277]

...

[Anonymous No.6658281]

,.,

[Anonymous No.6658321]

,,,,

[Anonymous No.6658323]

.....

[Anonymous No.6658326]

>>6658237
>I'm doctor loomis
kek was this intentional?

[Anonymous No.6658495]

>>6639298
>Modern AI is barely improving at this point, they pretty much stagnated, even GPT-4 is shit, and that's because they can't reach human levels, nor they will in the next few years or decades.
it's a gimmick, ai art is pressing the button 1000x times and getting deformed anomalies. even the ai is very fucking stupid and fails to grasp the most simple prompt.

ai ain't going to replace anyone in the next decade

[Anonymous No.6658498]

>>6640678
>What does it look like now?
like deformed shit

[Anonymous No.6658500]

>>6642762
>in a batch of 100000 deformed shit
wow

[Anonymous No.6658505]

>>6643064
>A: yeah you can within like 10 minutes of dialing in a prompt
lol no, gtfo faggot. ai doesn't even know what a human is

[Anonymous No.6658509]

>>6643079
yeah except that this is one in 1000000 images from batches full of deformed shit and ignored prompts. who are you fooling here?

[Anonymous No.6658560]

White thread

[Anonymous No.6658564]

.