Dancing Type




Dancing type is an exploration of audio-reactive, machine-learning-generated type varying on a 512-dimensional axis. 

Trained on a database of 4,000 fonts using a Stylegan 2 model, then exported using the open-sourced Lucid Sonic Dreams notebook.

Type unexpectedly morphed in ways not previously possible — from upper to lowercase, serif to sans, bold to italic, expanded to condensed, hairline to ultrablack — all transitioning smoothly without awkward in-between states.




This is typography that is not meant to be read; instead it is type that dances along to music, type that comes to life. It conveys aliveness, without having to actually use words — this type that is not a medium for a message but the message itself.



The following type is audio-reactive. Turn up your volume for the best experience.


The project was exhibited at the Weltformat Graphic Design Festival in Lucerne, Switzerland, and was selected to be part of the Type Director's Club (TDC) in 2022.
 


·    Machine-Learning    ·    2021    ·    MICA    ·    Created using StyleGan 2    ·    Tutorials by David Schultz    ·    Lucid Sonic Dreams Notebook by Michael Arafitz    ·    TDC Winner 2022     ·    Exhibited at  Weltformat Graphic Design Festival, Switzerland    ·    Featured by Contemporary Type