Creating Generative Art with the Subbody-Butoh-Method with text-to-image Neural Nets and Tatsumi Hijikata's Butoh-Fu...

a research project...

At the end of January 2022, my grand daugther Tala (17) showed me Wombo Dream, an app (and webapp) with a simple interface that lets anyone play with a simple text-to-image system.

Wombo Dream is a great (is't network knows a lot about Art and Visual Artists), it is easy to create something that looks appealing, but it was not creating art to my liking. Let's say it has a high kitsch output.

But it was a lot of fun, and invites co-creation and games. We played 'Guess the three words', and even more fun, every participant writes down one word, and we see what the combined spell (called the prompt) will magick up from the system. Try it and invent your own games...

I call the prompt a spell, because it is like a set of magick words. It took time and a lot of trial-and-error, to find out what combination of words seeds the creation that I was aiming for.

We (the AI and me) created more than 700 wombo's, some successful, most of them nice but boring.

See the 3 examples...

Soo, being bored with Wombo, I found another system, Nightcafe Studio. It gives better pictures, has a much better network, and playing with his I realized I was doing what I learned to do with my butoh students. Carefully observing their subconscious as expressed by the body, and tuning the words to get to some in interesting creation.

With the AI I was doing the same, and it stared to work for me. I found some spells to get into the darkness of the AI, it's subconscious.

And yes, feeding nightcafe Hijikata's butoh-fu creates interesting stuff