The reality-warping powers of AI have been getting busier over the past decade. We had 3D dioramas powered by computer vision; avant-garde style transfer; viral photorealistic selfie-tuning, selfie-retouching, face-swaps and -ofc-deepfakes; and a lot of frivolity (and hilarious) have fun with selfie filters (ohhai “Disneyfying” cartoon lens!) in between.
The AI-powered visual remix has shown, again and again, that it can grab attention. Though it might be harder to keep your eyes on once the novelty of an AI-generated effect wears off. (Self-editing apps don’t have this problem, beware; there’s a perpetual demand for machine learning as reality. booster seat.)
What is most remarkable about developments in AI-enabled synthetic media during this period is the speed at which these visual effects have caught on, aided by ever more powerful mobile processing hardware.
Wait times for a finished result can now be essentially instantaneous – a game-changer for producing (and potentially monetizing) the creativity and power of neural networks and Generative Adversarial Networks (GANs). Aka, the machine learning frameworks performing retouching, cropping or even generative modeling, jumping from a human prompt for their inspiration.
And while most app-based visual remixes of the past decade have focused on retouching/restyling/augmenting versus pure-play AI-powered image generation, that too is in changing.
Wombo, a Canadian startup that previously garnered attention for its eponymous AI-enabled lip-sync video app, recently launched another app, called Dream (iOS and Android), which uses AI to create original “works of art”, based on a text prompt.
That’s great, Great simple: you simply describe what you want it to paint – say “A terrifying tree” or “The worst sandwich in history” – choose a style from the selection provided (Mystical, Baroque, Fantasy Art, Steampunk, etc. ), or opt for “no style”; and press create.
Then, literally within seconds – I counted
You can’t even be bored during those few seconds of creation, because you get a glimpse of the AI at work: the app shows the rapid evolution of modeling – from starter marks, to a few blazingly fast additions that flesh out the Web. , to arrive, almost out of breath, at another finished composition.
Some of these generated artworks look quite impressive. Some… not so much.
But of course, two prompts do not generate the same image. So you can keep requesting a new image from the same prompt until you like the look of what you see.
In short, Christmas card artists and pulp fiction illustrators can probably retire now.
Everyone is an “artist” now.
That said, real artists should have less to worry about. Not least because art created by a human brain and body will only gain value once the world is flooded with “machine art”. (Just like every NFT minted dilutes the meaning of the term “digital art”…)
The quality of the “art” of the Dream app is certainly variable. Longer and more complex prompts seem to confuse it. So the quality of the output may depend on what you ask it to draw.
While his “style”, if it can be considered to have a unique style in the midst of so many pastiches, tends more towards the abstract and the distorted than towards the specific and the precise. Thus, portrait requests will not be rendered photorealistic. And it is generally more comfortable to represent the fantastic than the real. (A prompt “Madonna and Child” served closer work to an infamous failed Spanish church restoration than a crypto-Botticelli, for example.)
But the speed of production is impressive. Terrifying slash.
As soon as a new artwork pops up, the app wastes no time trying to sell it – displaying an option to “buy a print”, which links to its online store and looks like a nifty way to turn a visual tip into real income. (It offers “Custom Wombo Dream Print[s]which start at $20 for a matte poster or $45 for a framed print.)
If the startup can turn about 20 seconds of processing into $20+ in revenue, that could be a nice little money pipeline.
While most people have limited wall space on which to hang any kind of art, let alone images generated by, um, stupid machine – most of these random creations will remain firmly virtual. (“AI Art” might be perfect NFT fodder, though…)
Where “AI art” will fall in the fashion/cultural value stakes is certainly an interesting question to ponder.
It’s superior to clip art or stock photos, of course. And the Dream app release may also be more interesting than the average “art” print you might buy at Ikea. But the results can also be rather uncomfortable – or derivative – or non-existent – or just plain weird.
Well, is it art? Or is it just a visual output of a mathematical process? An abstraction of human creative skill that cannot convey real emotion or a sense of identity or soul because the code has none of those things? He just does what he’s told.
And you really want to hang a coded abstraction on your walls?
I mean, maybe? If it’s above all aesthetic. But hey, is it art or just wallpaper? Maybe Wombo should sell Dream AI wallpaper rolls or mousepads and t-shirt prints (merch), rather than “art” prints…
Lots to ponder.
Some things are clear: AI-generated art is incredibly fun to play. It’s a kind of visual catnip. A toy for the imagination.
It is also, without a doubt, here to stay. And AI models will continue to improve – depending on what we mean by “better” around such a subjective subject as art. (Perhaps generative art models will achieve better results by bringing the user more fully into the creative process – giving them tools to customize and manipulate machine outputs so they can to be [fine]listening until they are closer to what the person imagined, or else to feel more personally unique and meaningful. Or, in other words, a more hybrid creative process can yield more powerful and moving artistic results.)
There will also be dozens of these arty AIs, each producing different “flavors” and “characters” of visual output – derived from their training data. Or even, if you want, art AIs with different “styles”. (But maybe “specialties” is closer to the coded brand.)
There are a number of other GAN-based image generation AI tools – and I confess to being A big fan of Pixray’s system (the pixel art outputs are especially cute) although its processing speeds are much, much slower – but Wombo seems to have been the quickest to appify and monetize this technology.
The next decade of reality-bending machine learning will be quite the journey.