• Welcome to Autism Forums, a friendly forum to discuss Aspergers Syndrome, Autism, High Functioning Autism and related conditions.

    Your voice is missing! You will need to register to get access to the following site features:
    • Reply to discussions and create your own threads.
    • Our modern chat room. No add-ons or extensions required, just login and start chatting!
    • Private Member only forums for more serious discussions that you may wish to not have guests or search engines access to.
    • Your very own blog. Write about anything you like on your own individual blog.

    We hope to see you as a part of our community soon! Please also check us out @ https://www.twitter.com/aspiescentral

When A.I. Generated Art Becomes Something You Cannot Unsee

Pretty sure that video wasnt exactly taken seriously by whoever created it, and it's also 8 months old, using a tech that isnt exactly ready for the prime time. Even Midjourney doesnt yet offer a video-related function. I actually cant quite tell just which AI was used for it.

Here, I'll give a bit of contrast, with one I did myself:


That one was made using a combination of Kaiber (for the video) Midjourney (producing the image that I then fed into Kaiber), and the MJ creation is the result of a very, very long string of iterations that all began with one of my own drawings that I fed into it (and then later I used more drawings of mine to bend the style). The final result is pretty much what I wanted out of it, the original image I gave it reimagined into watercolor with flowing hearts and such. The only issue is the weird hand that appears for no reason for a second or two but I'm too lazy to go fix that, and the video is merely a test in any case. The video was also meant to work within Kaiber's current abilities, rather than trying to push it outside of what I know it to be able to do well. These things right now aint good at like, action and large amounts of movement. And it also depends on which AI is being used.

As it is that bloody video took a long time to do, and is built off a veritable archive of images that came before it, or were inserted into the process of getting the final image (which can be seen briefly at the start of the clip).

There's always more to the story, and things often aint as simple as they might first appear.

Artificial Stupidity Systems video deform test up next...
 
I need to see this AI movie.

I saw something interesting, someone made an AI generated version of Ingmar Bergman's 1966 movie "Persona" and called it "Another Persona". They have used AI to replace Liv Ullmann with another actress. And after watching the movie she said she was a little freaked out and she wasn't really sure how to feel about it.

 
I saw something interesting, someone made an AI generated version of Ingmar Bergman's 1966 movie "Persona" and called it "Another Persona". They have used AI to replace Liv Ullmann with another actress. And after watching the movie she said she was a little freaked out and she wasn't really sure how to feel about it.


Now someone just needs to write the utility to "grep" their entire movie library to replace all the action heroes with Pauly Shore.
 
I don't understand the point of AI "art". Why should I bother to look at something that someone else couldn't be bothered to create on their own? Yes, it will be used a lot in advertising, but I'm also great at avoiding looking at that.

It doesnt necessarily work the way you think it does. I know a lot of the anti-AI-art brigade likes to say that "oh you just put in a single sentence and you get a masterpiece" but that's not how it actually works. Unless you want the same nonsense in the same style with the same framing the AI always defaults to, that is. And of course it defaults to the same style if the user does that, they barely gave it anything at all to work with, didnt iterate, and didnt add more to the process.

As has been said elsewhere, REALLY using this stuff doesnt involve just a prompt. And it also isnt done in a vacuum.

Here, let me show you something:

sophie_carre__cb146dbe-a5e7-4223-a1a0-d2b169a2c3c7.png


There's no prompt here, not really. Didnt type a word.

This thing is an amalgamation of my own drawings (and I have way more than just this, in many different styles), and yes, it's meant to look jagged and creepy, that's the point.

I drew the things I fed into it myself... which took bloody ages (and I've shown them on the forum before), fed them into it, and basically just had it merge them. From there, iteration after iteration until I get something I like. During the iterating I can feed it other things of mine to steer the style.

This means that the final image took probably about 8 hours in total to produce. Only one of those hours involved AI in any capacity.

I can also do this with my fractal images (I dont have any examples of that at the moment).

On top of that, I can use these for further inspiration. Like, I look at that heart thing, and I think: Okay, but what if I made that, except out of watercolor or acrylic paints? And I let some of it just sort of drool down the paper a bit, like that part at the bottom? How would that look? Only one way to find out...


That's only my blends, mind you. A full on stable diffusion creation is different from that. That can involve: prompts that look like a giant paragraph of utter gibberish (tags and commands and such), photoshop, tablet drawing, renders, 5 bazillion iterations, merging, and... yeah this goes on, and on, and on. Someone really skilled at this can make things that I cant even come close to.

Making AI art by typing in a prompt and just hitting go is one thing. But actual artists using this arent going to do that, nor are they going to make it the entirety of the workflow. It'll simply be another piece of the process. How MUCH of that process will be up to the artist based on what they're trying to create.

Artificial Stupidity Systems video deform test up next...

Yeah... that's ALSO built out of my own drawings, so...
 
A person giving instructions to the artist is still not an artist. At best, they are the Art Director.

Art too will eventually become as contentious as the issue is presently pertaining to actors. I just see a lot of litigation ahead for so many things created so easily through the use of AI. Which well may proliferate Intellectual property rights suits on steroids.

Reminds me of the first time I saw actor Peter Cushing in a more recent Star Wars film. He had been dead for some time. With his survivors litigating accordingly.


 
Last edited:
A person giving instructions to the artist is still not an artist. At best, they are the Art Director.

You missed like 90% of what I said.

Think about it:

The heart there. The construction materials were entirely out of images I drew... by hand... each one taking about 4 hours. Two full images. Full art pieces. I drew both. By hand. From scratch.

I then took them and used AI to simply... merge them. There's no prompt, no special commands, nothing. And from there, I can iterate... without prompts. Or I can jam other images... which I also drew... into it to steer it further. But that part wasnt necessary. There's no need to do much beyond some mild iterations to undo quirks. I already did the real work.

With the AI only doing the amalgamate process, it only was involved in 1 out of the 9 hours required to make the thing (well, probably more like 20 minutes, I was distracted during the merge phase). Again, literally built out of pieces I created by hand. All the AI did was mash them together.

Considering all of that, tell me: Who made that heart?

Not quite that simple of an answer, is it? I find that this type of question gets different answers from different people, which should say something.

And am I an artist? Best be careful before answering THAT one. I've already shown things I've personally made on here many times... without AI. Generally, on actual paper (I specialize in lettering of all types and things like drawing plants and putting hearts on everything for some reason), I'm not a big fan of digital art outside of my fractals (and those are an entirely different and very bizarre skillset, I use a suite of about 12 separate specialized programs to make those). A local shop in my town has a rack selling cards made out of my drawings/paintings, even. Though I'm still not quite sure how THAT happened, but I digress...


I'm not a special use-case, either. The complex process isnt just me. MANY others do it this way, but can go way further. Like a professional artist could take it MUCH further. Drawings, paintings. 3D renderings (Blender). Photoshop, so... much... photoshop. Revisions, redraws, iterations, more and more and more, so very, very many tools and possibilities... if you actually watch someone do this, someone who REALLY knows what they're doing (and isnt just trying to demonstrate the AI), someone who is a professional artist, the actual AI activity is like maybe 10% of the work. If that. Merely a tool in the box, like any of the other stuff. Why? Because what artist wants to just have absolutely everything done FOR them? What fun would that be? If they wanted to be a "director" they'd have bloody well gone and done that instead of the bonkers amount of training and practice they likely had to do to become a professional artist.

And then there's AI types. Just because someone used AI, doesnt mean it's the AI you're thinking of. For instance, taking an image already made and doing a drastic resolution increase... AI is needed to do that, but it's a very specialized type. The image itself? Doesnt change. Only the resolution being upgraded. Yet still, "AI" is involved. That's only one type. But people hear that and make a bunch of frankly rather silly assumptions.


Have a look at these:

DogAndDeer.jpg


and:

GroveRoad1.jpg


AI was involved in both.

Tell me: Where and how was it involved? What exactly are these images?


Saying someone "isnt an artist" because they use AI in their workflow is a ridiculous blanket statement. Not too fond of blanket statements, gotta say. Particularly when you dont know A: how it was used, B: what function it served in the overall creation, or C: what skillset, background, career, or anything else about the person there may be. Heck, you dont even know how often they use it. Maybe they only barely involve it every now and then. You dont know without getting to know them. Making comments like that is pretty nasty, when you think about it.

I know Twitter is in full hate mode over AI art right now, because Twitter does that (and once they get bored of AI, they'll find something else to scream about). That dont mean that it at all works in the way they tell ya it does.

And here's one final question: take someone who is disabled, right? They've got all sorts of creative ideas in their head. Have had these ideas for years. But they cant hold a brush, a pen, stuff like that. It's not that they didnt learn to draw... it's that they physically CANT learn to draw.

So, someone hands them an image-gen AI and says: hey, this is a way you can create. This can take your ideas and bring them to life... all you need is words and a deep enough understanding of the tools to spin those words into shape. And the patience to pull it off.

Are you about to tell that person that they arent an artist?
 
Such questions may well be reserved for six to twelve jurors who likely will have very little technical knowledge of the processes involved. Where presumably clever legal representatives of a plaintiff and defendant will attempt to sell them the most convincing explanation to support their client. As is the case with most if not all intellectual property rights cases.

Not necessarily an objective matter of right and wrong, but rather which side puts forth the best argument. Of course like all civil wrongs, they only count when one party is willing to pursue (and bankroll) the matter in a civil court of law.

Which could likely reflect each and every case being litigated and adjudicated on their own individual merit, without any sweeping conclusions. Lots of headaches over such an issue.
 
Last edited:

New Threads

Top Bottom