Disco Diffusion, et al. Machine and me, secret city spots, hot air balloons and fireworks on our way to the circus. We’re still on the way. That first AI written and sung lyric you hear is “She laughs the sound of steam” and there is a lot more of that.
AI/ML generated animation with camera movement, better cohesion between frames, camera changes and prompts. Took about 36 hours of GPU time on a P100. Iterations per frame seem to depend on the frame and what’s scripted. Prompts were partially generated by a surprisingly poetic language model. More on that later. Disco diffusion, ViTB32 LAION 2b, ViTL14, ViTB32, RN50, no secondary model, custom settings, custom cut schedules and animation schedules. Working on better interfaces for the workflow. A lot of human intervention. No GPUs were injured. Just egos.
The Lo-Fi soundtrack is a mix of ML and human generated sounds along with some friendly, legal, licensed samples. Let’s go to the circus.
Please like and subscribe. I’d appreciate any feedback, criticism, encouragement and especially people subscribing. It’s nice to have motivation to create things.
A large number of human + machine images are posted to the Steampunk Circus on Twitter regularly.
Fun Fact: One of the first known uses of steam to do work was in entertainment millennia ago. The carneys of the day used compressed steam to “magically” levitate various light objects much to the amazement of the audiences of that era. Pretty wild! Very Steampunk! Dare I say, science is awesome! Like a graceful acrobat.
Stay perpetually preposterous at the world renowned Steampunk Circus!