previous up

Seminar: Between Images (2020)
Invisible Typography
Refael Blatt

Typography exists as a tool of communication and like every other tool we use for exchanging our thoughts, ideas and desires, its focus lies (most of the times) on the message, rather than showing off its own characteristics or visual appearance. As Adrian Frutiger wrote in his book Denken und Schaffen einer Typografie: »A good font is one that withdraws from the consciousness of the reader, to be the sole tool of the writer's mind and the reader's understanding« (Adrian Frutiger, ›Schriften‹, Birkhäuser Verlag, 2009).

For a letter to fulfil its job, it has to be ›invisible‹, meaning the properties like the different heights, widths, stem sizes and counters have to be in a perfect figure/ground balance with the other letters from its font family, as well as with the relationship of the platform which transports the letters. If a letter is not providing these essential characteristics, it will look off balance and the flow of the reader will be disrupted.

That lead me to the following question: Is there a bias in the actual design of letters? The purpose of this project was to look at the versatility of letters. By magnifying the tiny, by eye not recognizable details, the letterforms transformed into new abstract forms.

In order to have a big dataset, I downloaded the whole Google Fonts Collection2 and locally loaded and exported the same letter in white every single font into a black 1024×1024 px canvas using processing3. The collected dataset has then been trained with StyleGAN2 in RunwayML. The Artificial Intelligence then visualised a combined result of all the letters in a beautiful latent space walk. After creating various animations using p5.js5, I was able to use single frames of those newly generated animations to create interesting patterns and visual rhythms.

With this experiment, I visualised what makes a letter invisible. The change of a serif to non-serif, the various ink-traps of a letter were made visible in an animated version. What interested me the most, were the playful and unexpected connections the AI came up with.