This jam, I wanted to experiment with using an AI art generator (https://www.wombo.art/create). My normal collaborator (programmer @allthebees) wasn't available, so I wanted to try using an unconventional programme as an engine - and settled on using Indesign to create a hyperlinked pdf, and an interactive graphic novel.

Dream seemed very good at generating cityscapes, portraits, and pipes. With a bit of massaging in photoshop, it was possible to edit the results so that 'characters' appeared a tiny bit consistent, or to match colours between panels to appear on the same page. I wanted to try to build a narrative which dealt, at its core, with the way that AI art generators iterate, and produce multiple similar, but not identical outputs. In the available time, I didn't manage to finish a satisfying story arc: if I had infinite time, the narrative would deal with the characters realising the limits of their world (and lovingly pointing out some of the inconsistencies in cyberpunk/biopunk sci-fi...). All they know is that every ten seconds, another train arrives.

The thing that I managed to make in the time is really just a proof of concept - with time, I think it would be good to: * speed up the opening * increase the interactivity (would require some coding, but maybe have a 'blink' button, so the images on each 'page' change while the text stays the same?) * have a complete gameplay loop, where multiple choices lead to a good/bad/neutral ending * dial up the uncanny/horror/discomfort - e.g. having the pipework flicker into recognisable objects/animals

But I think there might be legs in exploring the idea further? The AI was very good at generating: * cyberpunk cityscapes * portraits in the styles of specific artists (here, portraits like Egon Schiele, pipework like Terry Gilliam's Brazil) * pipework. So, so much pipework.

I'm not sure how I feel about AI art generators, but it was a fun weekend.