Artistic Intelligence

Yes, artistic, not artificial. The emergence of AI to the broad public halfway 2023 through chatGPT has opened many eyes… and many doors towards unforeseen possibilities. Right after the first text generators came image generators like Stable Diffusion.

Although the first image generators were already astonishing but low in quality, the quality has made leaps in quality and prompt understanding. Although logically one would think that AI would be genius in logical constructions and technical depictions and would completely lack understanding of ambience and emotions, it turns out to be the other way around. Prompts that ask for “happy” or “fearful” turn out images that beautifully showcase these emotions. Depicting humans with the right amount of fingers or just one knee in the middle of a leg seems to be harder to understand for this computed intelligence. This is exactly opposite of what one would expect from computers making images.

How AI changed visualising…

And that is what turns out to be the perfect tool when visualizing ideas. I use AI for imaging, as a tool, but not necessarily for creating final results.Imagine a poster for a spaghetti sauce brand. Of course you go for the, maybe cliche, Italian background of green and yellow curvy hills with pointy trees and flat terracotta roof tiled mansion. You look up pictures on the internet for reference and combine the best elements to become your first and probably definite outline for your composition.

With AI, prompting the description as stated above, you get a whole series of images instantly. Another click on the button generates another series of similar images in slightly different compositions. This way you create a closer reference than you would otherwise be looking for on the internet the whole afternoon.

Better yet, AI visualizers have a slider that gives generators less or more space to be creative. Sometimes, in a series of generated images, there is that one image that stubbornly goes off the tracks and portrays something you were not looking for but fits the original concept even better. To stay in the spaghetti sauce narrative: imagine there is one image of the cliche curvy hills, but this time at night with beautiful lights emitting from the mansion. This fits the original concept even better: it’s dinner time, festive with all those lights and still really Italian. A new angle you were not looking for, but it improves the image and whole feeling of the campaign.

Will this take out the designer as a middle man? Not as long as this campaign image needs the output in different formats (poster, flyer, website imagery, packaging). Will it be faster? Probably some… but the giant leap that is being made, if you ask me, is in the concept and quality of the output. Pixar quality is now available by prompt. Custom or stock photography is now a few prompts and clicks away and no money(photography trip to Italian curvy hills) is wasted if you try a different angle.

… and we haven’t really started yet

New innovations (yes, it’s double and rightfully so) arrive weekly. Last week's stunningly improbable innovations are outdated by this week's innovations that blow minds on a nuclear level. Next week, when you have just wrapped your brain around the possibilities, the mind will probably be blown again on a supernova scale.

Developers are racing towards an undefined point on an undefined horizon and barely have time to actualize a monetization strategy for their services. And that’s great, for us, for now.The quality of generated images and the understanding of prompts improve daily. This now moves towards moving images, generated talking humans(generated text, generated voice, generated face, generated movement, in a generated background) that can no longer be distinguished from real humans.

In visualizing this will mean that, instead of just an image, the festive nighttime spaghetti meal in the curvy Italian hills may as well be a commercial or a movie… featuring Beyonce, The Minions and your mother in law.

Where will it end?

I’m expecting PromptFlix within a couple of months, where you define the contents of your custom generated series. And after that? Total Recall or The Matrix. Will designers still be needed? Possibly not, but they will probably be the ones with the prettiest Artificial Realities.