AI art criticism general - The only place on the internet for right wing criticism for AI art.

[Smack Patrick]

I feel indifferent towards AI art

In my opinion (artfag). There's no sense being a luddite when it comes to AI art. AI has been a godsend for furthering my craft since I can use it to generate references I can draw from. No more having to browse through a bunch of different image hosting sites for references when I can load up the AI generator and create the references I want.

I view AI Art like I view taking steroids at the gym.
If you want to go ahead and take steroids, do it. I don't care, I'm doing my own thing.
Just don't try to compete in competitions labelled "natty only", don't lie and tell me that you're natty, and do not try and bullshit me that you put in just as much effort as people who are natty.

AI has genuinely mind broken so many artists it's crazy. The art community online has broken into the following 3 core groups:

1. People using AI, be it for memes, drawings, etc.
2. People who used to do art, but now all they do is whine about AI. (Had to unfollow a lot of people because it was getting fucking annoying.)
3. People who do non-AI art, they have their own opinions on it; positive or negative; but they've got more important things to do than whine. Like drawing.

If you're an artist and your Xitter timeline has more posts dedicated to advocating against or complaining about AI art than it has posts of your own art. You're not gonna make it.
Stop whining, and just draw. Simple as.

 

[John Helldiver]

In my opinion (artfag). There's no sense being a luddite when it comes to AI art. AI has been a godsend for furthering my craft since I can use it to generate references I can draw from. No more having to browse through a bunch of different image hosting sites for references when I can load up the AI generator and create the references I want.

I view AI Art like I view taking steroids at the gym.
If you want to go ahead and take steroids, do it. I don't care, I'm doing my own thing.
Just don't try to compete in competitions labelled "natty only", don't lie and tell me that you're natty, and do not try and bullshit me that you put in just as much effort as people who are natty.

AI has genuinely mind broken so many artists it's crazy. The art community online has broken into the following 3 core groups:

1. People using AI, be it for memes, drawings, etc.
2. People who used to do art, but now all they do is whine about AI. (Had to unfollow a lot of people because it was getting fucking annoying.)
3. People who do non-AI art, they have their own opinions on it; positive or negative; but they've got more important things to do than whine. Like drawing.

If you're an artist and your Xitter timeline has more posts dedicated to advocating against or complaining about AI art than it has posts of your own art. You're not gonna make it.
Stop whining, and just draw. Simple as.

well, in my opinion, the actual problem is the way it fundamentally works and what that means for IP law if cockeaters like elon musk get their way on the issue.

the fact is that the way it works is illegal, they know it's illegal, they know nobody can "prove" it but everyone knows.

for profit, of course, i can't believe some out there seriously believe the giant gigamegagovernment corporation is their friend.

 

[The Wise Shaman]

While AI does have the potential to be a good tool for making visual arts, in its current state, it's rather limited. Creating a full image or a video in one go makes it really hard to make any further adjustments to it. The only way to use such AI to make good art is to plan around this limitation. I've seen an artist on Pixiv (I forgot his name) who makes fan art for a 3D cartoon with AI by training it using that show, generating each part of the image one-by-one, making adjustments to them, and then editing them together to make the full image.



Imagine something like this, but all assets are AI-generated and it looks coherent.

If we want to make an AI model more useful artistically, then I suggest creating one that can use the same software to make art as we do. This would make it way more flexible and adjustable. Without this, artistic application of AI image generation is basically limited to shitposting, making mediocre-quality porn for degenerate coomers, and replacing stock images or creating really specific ones.



Now let's mention AI slop, as it is very relevant when discussing AI art nowadays.

I think when we mention slop, we shouldn't forget that the enshittification of the internet began way earlier than the rise of AI text and image generation. People have been putting out garbage way before. Just look at kids' Youtube. The rise of AI didn't stop slop channels from producing "pregnant Sonic shits and dies" videos using 2D or 3D animation software. Elsagate was in 2017 for God's sake! The only thing AI did is it added a fire to the fuel, and it added a lot. Cracking down on AI slop is only possible through cracking down on all slop, which can only be done by making it unprofitable. Let's just say social media giants don't really want to lose money, so it's unlikely. The only thing we can do is segregate ourselves from the wider internet into smaller communities, basically returning to the good old days of the internet if we want to free ourselves from the slop.

 

[fatrickrapedme]

being a luddite

As someone that actually likes and enjoys art for what it is this argument is so stupid. You are not a "luddite" for enjoying art and wanting art to still exist. AI "art" isnt art. Art is expression. AI is anti-expression.

 

[Slurred]

If we want to make an AI model more useful artistically, then I suggest creating one that can use the same software to make art as we do. This would make it way more flexible and adjustable. Without this, artistic application of AI image generation is basically limited to shitposting, making mediocre-quality porn for degenerate coomers, and replacing stock images or creating really specific ones.

This already exists in the form of inpainting, img2img and controlnets. You can get an enormous amount of control over an image by regenerating parts of a given image at differing denoising strengths, using your own manual detailing as a guide for what to change. Slop merchants don't do any refinement of their images, and editing isn't front-and-center in the most commercial/simple image generation tools, but these techniques are available and honestly indispensable for anyone serious about using AI image gen to execute on a specific vision.

The best tool I know for doing this sort of thing is Krita AI Diffusion, a plugin for the Krita painting program. Some videos about it:

 

[Squawking Macaw]

My biggest gripe with AI is that it's ramping up the amount of slop being dumped on the internet. Yeah sure, it was already pretty bad before, but AI has somehow made the problem worse. It's shitting up google images and making it a pain to get real references, you'll be flooded with shitty AI images that don't even have the decency of looking like the thing you're trying to get a reference of.
At least you can put before:2022 at the end of your search to get rid of them, but that also gets rid of anything recent.

 

[Cnidarian]

There is a certain individuation taken out in the AI images as well. For example, the AI images there is an orange tint that isn't present in authentic ghibli movie screenshots. The reason is likely the fact that alot of artists give there works an orange tint and alot of people like orange unknowingly and associate it with the cozy feeling of ghibli. So all of the false ghibli images are stripped of there individuality unknowingly and are mixed in with everything else even if that doesn't seem like the case.

I think it's a universal problem with chatGPT itself turning everything sepia for unknown reasons.

 

[Nip Slip]

My biggest gripe with AI is that it's ramping up the amount of slop being dumped on the internet. Yeah sure, it was already pretty bad before, but AI has somehow made the problem worse. It's shitting up google images and making it a pain to get real references, you'll be flooded with shitty AI images that don't even have the decency of looking like the thing you're trying to get a reference of.
At least you can put before:2022 at the end of your search to get rid of them, but that also gets rid of anything recent.

It's mainly because of third worlders.
They build digital guides for one another on how to abuse social media/website payouts in driving engagement strictly using AI. As if polluting their own rivers wasn't enough, they're now e-littering slop everywhere online for a literal living. Video content, incoherent blogs and "sources", 2d and 3d assets, stock images, social media group pages, you name it.
What irks me the most is when they sell their prompts to the public on websites like Etsy or even Amazon without disclaimer that the product wasn't made by a creative.

 

[Mage Killer]

I think it's a universal problem with chatGPT itself turning everything sepia for unknown reasons.

Because humans naturally like the color orange, because alot of paintings take place outdoors under sunlight, or indoors under torchlight, it caused chatgpt to just give everything a warm tone, and AI prompters because of their emotional shallowness simply don't care.