machine learned typography
To explore new letterforms I trained a StyleGAN2-model with images of typography that I took around my neighborhood. These include signs, clothing-tags, graffiti etc. For further exploration I used SinGAN to generate variants of the resulting letters and Glyphs to test turning them into a font.The results surprise with organic forms and vary greatly in their level of abstraction. They span from rather conventional to quite exceptional letterforms.
