If Jodorowsky directed TRON

Here's the latest issue of GOOD INTERNET-newsletter, published first on Sustack (Öffnet in neuem Fenster).

For a long time I wanted to post a list of my favorite stuff done with image synthesis, but its a bit of a futile endeavour: The stuff posted in the subreddits is a flood of shimmering light, most of it is technically impressive and some of it is actually clever, and the top percentile of that crosssection still is way too much to filter, so i just postponed it and let the images wash over me. Maybe I’ll come back to it, maybe i’ll do it regularly, most likely i’ll incorporate the most mindblowing stuff in the usual GOOD INTERNET-issues, and rarely, i’ll devote a single issue to one amazing latent space exploration like i’m doing now.

With Midjourney v4 the jump in quality of AI creation is remarkable. Photorealistic faces combined with extremely detailed artwork while still preserving an edge of dreaminess which gives AI artworks their unique feel (besides the mutant hands ofcourse, which even Midjourney v4 wasn’t been able to fix). The results are mindblowing. Treat yourself to the stuff happening in the Midjourney-sub (Öffnet in neuem Fenster) or at the hashtag (Öffnet in neuem Fenster) on the tweeties. And mind you, this is not necessarily about art. Ofcourse these machines can create art, but that’s beside the point. You are looking into parallel universes where Charlie Chaplin played a robot in Westworld, painted in the style of Van Gogh. These things are not art, these are stochastic variations of our past, and we can create any combination of our combined knowledge.

For instance, this fantastic imagination posted to the Midjouney Group on Facebook (Öffnet in neuem Fenster) by Johnny Darrell: If Jodorowsky directed TRON (Öffnet in neuem Fenster). This is amazing stuff, and your head starts spinning if you look at it and I'm jealous of my kids who one day, when they’re old, will be able to watch this movie in a CKPT-file of the future.

Even today you could generate a wonky 3D-model from these Jodorowsky/Tron-images, and make it dance with EDGE (Öffnet in neuem Fenster) to a soundtrack by JukeboxAI. Soon, you can feed a storyboard and a script, and it spits out a movie and you may be able to sell the accompanying AI-model as newschool merchandise: A mobile themepark about your movie, ready to be explored by everyone. Generative audio for these movies will be tricky, because voice and ambient sounds need to be not only in synch with movements, but some soundtrack also needs to be engaging, fun, climactic and nice to listen to. Because these details are hard, i guess this tech is still far away, maybe 10 years, maybe longer. But we’ll get there, i have no doubt about that.

For us, stuck in the past, we can only look at images from these parallel universes our grandkids will immerse in. I’m sure this technology and it’s sheer endless escapist possibilites will have an intense impact on our psychology and cognition and not all of them will be good.

But damn. Look at this: