Leonardo.ai has this new-ish feature where you can draw something on a canvas and it'll re-generate it into proper art. It's pretty cool although I know a lot of other platforms have been capable of it for a while now.
In this case though, messing about with it seems to be free and unlimited on Leonardo unless you wanna start upscaling and such.
Now for the most part, I'm quite anti-AI but also acknowledge its inevitability.
It sucks for music, and it's starting to suck even worse for artists. I even feel the psychological effects within myself. I'm just not really impressed by award-winning photography anymore. I remind myself of the effort and time that goes into them but it somehow just doesn't hit home as hard as it used to when I genuinely have no idea what's real anymore. Like, I literally can't tell world class photography that took years to make, from a 3-word prompt on Discord.
But with all that in mind, I think there's stll some... wholesome purpose to be had with AI. I think it's still a powerful tool for inspiration for example, among other things.
So, I'm playing around on Leonardo with the idea of perhaps creating blueprints of images for a story I wrote as a child, and couldn't help but attempt to confuse it as much as I possibly could to see what wacky results came out - certainly, no human would come to the same conclusions.
I wrote a prompt like, say, 'Old Dog', but then I drew a reference image completely unrelated to the prompt, such as some random music notation.
You can move the slider left and right to determine how much creative strength to give the AI, and I must admit it's fascinating watching it seamlessly process the two into super-creepy, warped results.
Musical Dog
For this example, I've set to 'Cinematic' mode.
So this is full-strength creativity from Leonardo. You can see it's clearly got an old dog. No problem. It also has the colour scheme in the frame and interestingly, the top left--> bottom right trajectory of the musical notes.
This is a little over 50% strength, and you can really start feeling freaked out already. BUT, I'm starting to see the way it's forcing the notation reference into the image, treating them as legs for the most part - genius.
Yeah... Or, I can switch it to a little friendlier 'Anime' mode and get this rather musical-sighted dog:
Much cooler! Here'e 'Fantasy art' mode:
Tree Whale
I'm struggling to see the connection between full creative strength and my beautiful tree, other than how in-your-face it is:
Lowered to about 75% and you can definitely see where it's coming from, but the addition of a pensive looking man is a strange choice. Perhaps like the movie the whale, it's some kind of deep metaphor of one's own reflection.
I really like this next one. They did away with the Blue in my prompt 'Blue Whale' but then replaced the tree trunk with some blue... something. It looks like something from Spongebob XD
Finally, here's about 20% in photography mode and is truly creepy:
I have no words to describe this at risk of thinking about it enough to get nightmares later.
Inspired?
Yeah I'm not sure my prompt conflicts specifically were particularly inspiring. But just to justify this whole system, here's something cool you can make if you're a bit more sensible, here's about 70%:
Tree Whale 😆😆😆
Have been playing a little with Leonardo as well.
Tons of possibilities, for sure.
So far wasn't able to get something cool, though got Ernie in nature with a flower. Als Biden en Trump holding each other's hands. The latter I was hoping to get tons of reaction on X-platform, but my post got snowed under I suppose.
Didn't know about the self-drawing feature. Will have to try.
I feel like Leonardo still struggles with like, limbs and basic stuff I assume Midjourney and the like have figured out. But aesthetically it's a lot nicer, and the community postings show that you CAN get excellend renditions if you know how... I'm just too casual a user to bother... at least for now!
A reason why I started to play around a little with Leonardo and left Midjourney for what it is. Because the results are nicer, as you stated.
Limbs, fingers, something Leonardo goes in overdrive. 6 fingers per hand is easy. Or two books in hand when I don't prompt the engineer good enough. Somehow I feel Leonardo needs all words in one sentence. Quite frequently it doesn't take into account whatever I write down in subsequent sentences.
The two mentioned creations I kinda like.
Haha excellent. The innocent and wholesome smile on Biden is golden