Parkour civilization
You got the name wrong
oh yeah
i’d think watching all of skibidi toilet would’ve been the thematically funnier thing to bring about that change in you
I never found time to respond to you contemporaneously and then it felt weird to dig it back up so I apologize I never responded to you in kind, but this is hard to believe because each image starts from a field of noise and is fundamentally a denoising process. You sort of fundamentally cannot have that fine of control because from run to run it’s starting from an entirely distinct noise map. If you are getting minute changes from run to run, that likely means that the prompt is so specific that it’s essentially plagiarizing an existing work.
@Aleph Here’s the links I was showing of people making Thoughtful Art with AI - the first one is somebody who makes comics with a combination of Photoshop and AI generation, and the second one is somebody just making an image with pure AI, starting with a specific vision that takes them like an hour and half to get quite right.
Hearing people talk about this is a big part of what turned me on AI art Being Art, it’s not just a slot machine that you get images out of, there’s real ways you can finely control what you’re getting. Less so than normal art, it’s got fewer decisions per square cm, but it’s still got decisions.
i think i never matured as a teenager did i. like mentally am i still there
i’m still not entirely convinced you’re not lying about being 25
in 2 months i’ll be 27
i…
what am i doing …
If someone is also getting a consistent style of artwork out, that’s also a good indicator that they are, unintentionally or otherwise, providing prompts that winnow down the data set to a particular artist or small number of artists, and that the output is heavily derivative of that artist’s. The part that sucks is you could generate those images without even knowing and that are different enough to not be reverse image searchable, but similar enough that you could identify the source of the latents if you looked through the original artist’s portfolio.
Nah, there’s technology! AI’s not just “run a prompt and get an image back” - this is covered with examples in the links I just posted, but a quick summary:
- You can run “variations” on a specific image you get and get similar but slightly different images back, allowing you to take an image you like and get random slight changes until you find osmething you like
- You can prompt with image-to-image, allowing you to, say, put in a previously designed character and “ask” for that character doing something else - the Krita AI plugin I was messing with earlier even just gives you a little pose skeleton you can physically move around to change the pose a character in an existing image is in
- Similar to the above, you can “ask” for a modification to an existing image. This has been around for ages, it’s one of the htings DALL-E 2 was bragging about: take an image, select a part of it, go “remove the flamingo”, and boom, the flamingo is gone
- You can use image-to-image to turn a rough drawing into a more rendered one, creating a sort of blend between AI and traditional art - you block out the shapes, it creates an image using those shapes
Just the other day I heard about an account on Bluesky that claimed to be an artist but was really using AI and clearly prompting in a way to ripoff another artist’s style and they were finding success. Like that fucking sucks and I don’t care if they spent an hour and a half making each image look good if it’s just splicing the work of a particular artist who spent like 10-20 hours on the fully rendered peace and years of practice.
No, not really? There are thousands of impressionists out there, there are thousands of people doing pencil drawings, there are thousands of uninspired anime girl artists, there’s more than enough out there to get a consistent style without “ripping off” a single individual artist. I feel like you’re underestimating the volume of training data here.
And one person ripping somebody else off does not destroy the value of an entire medium. Even a lot of people ripping others off does not destroy the value of an entire medium! Does AI make it easier to rip somebody off? Sure. But digital art made it way easier to trace or directly steal art, that doesn’t mean it’s not art. You can be derivative in any medium; if that particular artist on Bluesky didn’t have access to AI technology, there’s a solid chance they’d just trace other people’s art, they’d be derivative anyway. It’s just a new way to be derivative, it’s just easier. Technology enables bad things sometimes, it sucks, doesn’t mean the technology itself is poison.
have you considered that since the printing press enables easier plagiarism we should return to stone tablets
we need a 1v50 thread for this specific reason
atlas vs 50 on “is ±0 different number”
litten vs 50 on “is it lying if you unintentionally tell a falsehood”
may vs 50 on “is ai art real art”
tutuu vs 50 on “is tutuu sexist”
[Edit: these two paragraphs are separate points and not really related/flowing into each other]
A lot of people talking about AI go “here’s this specific situation, it sucks; AI therefore sucks”. Here’s this company that laid off people to replace them with AI and the art got worse, here’s this person who plagarised somebody else’s work using AI. But that’s just… throwing a bunch of bad situations in a pot and turning them into a worldview.
//
It is possible to be original with AI art, because creativity in art comes from the combination of novel concepts and techniques: you are never going to be the first person to paint with a particular medium, you are never going to be the first person to use a specific colour palette, you are never going to be the first person to paint an object, but you could be the first person to oil paint a half-eaten apple in neon colours floating in space. That’s how you use this technology to make new things.
its if 0 has a -0 or a +0 sign at all!!
man who could possibly have seen the fact that me bringing this up would start an argument coming