Parametric Typography

Like my previous post on the history of parametric design, this post is another B-side from my PhD thesis. It considers how parametric modelling might inform typography, which was an idea I explored in my thesis but eventually discarded.

For two years I studied how parametric design was impacting the practice of architecture. I had three case studies, a number of smaller experiments, background literature, and the official blessing to being writing my thesis. At this stage I did what most people do when asked to write what is essentially an 80,000 word essay: I began procrastinating.

I began developing a parametric model to layout my thesis, a project I justified by calling it my ‘forth case study’. The first three case studies explored parametric modelling in the context of architecture, this forth case study was to extend the argument into a wider practice of design. The thesis itself would become supporting evidence for the text within. Since much of my thesis assessed how parametric model flexibility was impacting the practice of architecture, I was interested in what this flexibility would do to graphic design. Could the layout begin before I finished writing? Could I make last minute changes to both the text and the design without either being disruptive? Could the medium and the message co-evolve together? What would this look like?

Visually I was inspired by the initials and embellishments that Wired magazine commissions from designers. Each article has a unique feel but there are underlying rules that make the overall issue feel cohesive. I also found inspiration in an number of logos that exhibit a similar parametric quality (see images below). The City of Melbourne (where I currently live) has a “flexible logo”, which is the silhouette of a three-dimensional ‘M’ overlaid with patterns that change based on context. MIT Media Lab has forty-thousand computationally generated logos, each thematically similar but visually different. And Pepsi’s new logo (2008) derives from a set of geometric rules that enable each product to have a different permutation of the logo. I find these examples curious because brands typically become recognisable through repetitive exposure. Companies go to great lengths to ensure a brand’s colours, spacing, silhouettes, and sizes are always the same in order to improve the likelihood a customer will identify them. These examples seem to flip this thinking on its head. Jose Salmeron, writing in Smashing Magazine, calls them “fluid brand identities”. They embrace visual variation and find paradoxically stable identities within it.

Inspired, I began designing my thesis with a parametric model. Instead of the rigid consistency of Indesign templates there would be computationally generated layouts that were unique on every page. These would be unified through a system of proportions and gridlines. There would be a single font used in the entire thesis. Rather than the font being an outline filled with colour it would be the input for various parametric models. Each chapter having a model that interpreted the font differently; each title, each drop-cap slightly unique. (You can see the results at the end of the post).

I worked on this when I lacked the inspiration to write (which was quite frequently). I ended up developing a fairly elaborate system in Java/Processing that would read text files from Scrivener, generate PDF’s of all the title pages and other embellishments, before linking everything together in an Indesign XML file.

I was in love with the system I had created. My infatuation was so strong that I couldn’t think critically. When I showed my supervisors I thought we would have a long discussion about the ramifications for the graphic design industry. Instead they told me it looked overdone and childish. Initially I thought they were concerned that I had spent more time writing code than writing prose, but after a few more months of stubborn persistence I came around to their view.

Visually it is too much. The graphics overpower the text, the message lost in the medium. Worse still, the parametric models were this pastiche of common parametric tropes. While my text tried to elevate the academic discussion of parametric modelling, the design was essentially parametricism for books. I started over. My thesis now looks much like any other but below you can see some designs that never were. I have also included the Processing source code for the fonts from each chapter (Creative Commons, attribution). Perhaps you can find a use for them. I still like the idea and even if the designs don’t intellectually meet my aspirations they have something that I find visually compelling.

It wasn’t a wasted exercise though. In another post I will explain how I used these computational techniques to layout my much more conventional final thesis. And the logo for this site, the colours, the drop-caps, and the fonts, all derive in some way from these experiments.

Download the Processing files

Chapter 0

Chapter 1

Chapter 2

Chapter 4

Chapter 8

Chapter 9

Sample Spreads

Further Resources

• Flickr Generative Type Group 
• Reza’s Typography Experiments
• Generative Typography on Open Processing
• The Meta Font (thanks Tomás for suggesting this in the comments)
• Avería – the average font (thanks to Benjamin)
• The Letter Spirit Project (thanks to BitCraftLab)
• The Whitney Museum in NYC has a responsive ‘W’ logo