AI AI AI!
ay ay ay (Interjection): Used to express surprise or dismay.
In my previous post, which I have just learned I can apparently embed like this:
Dreamers of Dreamons (An Experiment)
My parents were away on yet another of their holidays last week, and I was tasked with looking after the fussy, needy, poorly-trained dog. As with the previous times this has happened, I used the opportunity to get away from my PC for a bit, because it’d be cruelty to expect the poor animal to spend all day in the prison cell I spend all my own time in.
(And yet Patreon STILL has no ability to do ANY kind of formatting on inline images, pffff!)
IN THAT POST, I said I’d be revealing the revised and additional Dreamon characters over the course of this week on my Patreon… which I can’t embed like that. Oh well!
I’ve written the first post there now, revealing Spryad and her dreamer. So go over there to read more about her if you’re one of my patrons! DO IT NOW!! (Since apparently including a ‘call to action’ leads to more engagement…)
I don’t intend to do this for all of them, but I also want to share that art here:
The reason being because a new and improved AI image generation model came out the other day, and Reddit’s erupted into a trend of ghiblifying people and memes with it. Y’know. Ghiblifying.
I’d also seen people asking it to turn their own character drawings into realistic versions, with startlingly impressive results.
Curious, I provided ChatGPT - just the free version - with the above image and asked “Could you turn these concept drawings of characters into real people?”
It actually asked for clarification first: “I can generate realistic interpretations of these characters as real people! Do you want them to retain their fantasy elements (like the fairy wings and plant-like features), or should they be fully humanized while keeping their essence?”
So I said “I want you to generate an image that maintains all the details of those characters - including their poses - but rendered in a realistic way.”
It then took about a minute to produce this:
God. I’ve been following the development of AI image generation since its inception, and this is maybe the first time a result has felt like a spear through my chest. It took me a long time to even be able to look at that image more than glancingly.
I was also impressed by this AI generation of Spryad from last July, from a text prompt describing her features:
That’s already way better than most skilled artists could produce - and we could never produce it as fast as the AI did - but the new one is far more impressive. Especially going from initial input art, meaning it had to interpret that image before creating its own…
It even got the awkward, geeky look of the dreamer (the one on the left) accurate rather than making her into some hyper-attractive model or something! Plus it interpreted shapes as things like leaves or fairy wings even though the drawings were essentially abstract symbols of them rather than accurate representations (though it interpreted the sprout things at the back of their heads - meant to be cowlick things, part of the hair - as actual leaves).
A lot of online artists dismiss AI ‘art’ as soulless slop, devoid of the ‘feeling’ they ostensibly put into their own work. They say human creations have inimitable appeal, that these soulless machines could never compete.
I fear AI art because it’s damn good, and only getting better…
(I think I’ve said all this a bunch of times in past posts, though each new development brings all the feelings back…)
This short video talks about the new AI model a bit, and at the end it briefly shows some state-of-the-art generations from a couple of years ago. I remember that, when they just looked like mangled messes of random parts… It’s evolving so fast.
My future feels hopeless enough as it is considering how I’ve spent my life hiding from, well, life. All the skills I’ve developed over the years of isolation will soon be obsolete, if they’re not already. I’ve no idea how anyone’s meant to decide on a path forward at this point. Nothing seems secure.
I would like to finish Dreamons before the world falls apart, though. And my website. I’ll get back to those tomorrow.
I'm surprised there isn't AI art of Mardek as far as I know. But now I fear I just opened a rabbit hole just by saying that. It's becoming increasingly more complicated of an issue as time goes on. Makes you wonder when we reach the point of singularity...but then again, we'll probably notice that point almost instantly.
"I fear AI art because it’s damn good, and only getting better…"
I agree. I want to say it's still largely lacking, such that someone just dinking around with it while having zero artistic inclination is unlikely to get much. Or what they get will have glaring flaws that perhaps they won't be able to fix.
But I imagined artists must have been upset when photography came around and people were able to snap realist pictures in a matter of minutes rather than hiring a portrait artist to render them by hand for hours on end. And back then, artists were still largely starving just like they are today.
In my personal opinion, AI is going to be a great tool for everyone, but also for artists. Just like serious photography is generally done by people with artistic visions and inclinations, so will AI art.
And for that matter, the ability to create high quality and impactful art in less time just means that there is more of it to distribute. Same idea with prints. It cheapens the art value in a sense, but it makes the opportunity for much grander creation, and I'm sure we'll be seeing that soon.