Propopsal for machine-learned typeface.

Role: Art Direction, Typography, Machine Learning

Neuro Gothica is a proposal for an A.I. font that is created by a neural network (GAN; generative adversarial network being trained on 300 different specimens of gothic blackletter fonts. The morphing movement we see is essentially a visualisation of how a neural network would attempt to recreate its own letterforms.

In an ideal world you’d be able to create an interface that lets you manipulate the forms by applying a blanket vector over all the letters so they morph in a consistent style. You could have coordinates within multiple dimensions and plugging in the same coordinates of the g to the a would essentially make them match in their typeface ‘family’. I still have no idea if this is even possible or how it would be put together plus you’d have to train each letter separately with labels which would take a very long time and render power, but an idea I’m still open to!