Sorry Dieter 29 Sep 2022
I was playing with Stable Diffusion to make ‘things in the shape of letters’. One thing led to another and in a couple of hours I’d generated a full alphabet of Dieter Rams-ian industrial design. So I selected a couple of hundred images and stuck them together. There’s a short (15s) video over on Vimeo.
This is how they come out of my home PC ‘raw’. I haven’t tweaked them, post-processed them, iterrated them, or used AI-upscalers or enhancers.
It was just a quick experiment really. To see how much you can extract the influence of a particular artist / designer / creative within a generalised data-set. And then use the model to bend the work into someting ‘new’.
I have all kinds of conflicting feelings about it.
But as a creative who has ‘control’ it’s pretty thrilling. So is spraying a flame-thrower around; until you realise what you’ve done 🔥😿.
It’s messed-up that the world’s creativity is being hoovered up, fed to a computer, having interfaces put on it, and being flogged as a subscription for $9.99. And Dieter will never see a penny. This is one of those ‘disruptions’ that (left unchecked) is going to be very profitable for a few people and potentially disasterous for many more.
On the flipside – my inner optmist sees amazing potential for artists and technologists to work together to create mind-blowing systems and tools and experiences. In ways that aren’t trying to replace the creators, but to give them ways to extend their practice and reward them for their contributions.
There’s all kinds of issues to solve, but I love the idea of co-operative GANs that reward artists for their work, ideas and inspiration flowing through a neural network.
That may sound like sci-fi nonsense, but this stuff is getting better and better at an alarming rate. It would be easy to dismiss this as visual novelty, as yet another wave of Google Colabs hit Reddit, spawning thousands of hyper-realistic warrior princesses ‘trending on Artstation’. But that’s like judging the future potential of the Internet by looking at GeoCities.
I’m pretty confident that within a couple of years you’ll be able to generate images indistinguishable from stock photos. And within 3 or 4 years we’ll be able to get to convincing totally believable video clips of the same quality. Which is going to have huge implications for all kinds of industries.
If you’re someone in the creative world who likes surfing the waves of what’s coming next; it’s time to start paddling…
Hey Computer, give me: ‘a bunch of dragons riding the waves like horses in that guiness advert featuring a track that sounds like Orbital featuring Brian Harvey’ 30 seconds. Oh and can you make a few different versions to appeal to different audiences…’.
If you’re interested in the topic Chris Boyle just sent me a link to this brilliant piece from the ever-interesting David O’Reilly.