Information Theoretic View of Font Design

dpmellis

Partly for my own curiosity and partly for a data visualization class I am taking, I wrote a python program to extract information about a wide range of fonts with regular and bold weights, in Roman and Italic. Mostly I use the number of on-curve points as a metric of glyph complexity, but I also count the number of smooth points that aren't extrema for added interest.

The data isn't super interesting from an overview perspective—it shows that serif fonts have more points than sans serif ones, wow!—but with a bigger sample of fonts, it could be interesting to identify outliers that would be interesting to examine in detail. 

The main dashboard is here, but there are other tabs to poke around in.
https://public.tableau.com/app/profile/daniel.mellis/viz/FontComplexityAnalysis/Dashboard1?publish=yes

The code for extracting the data is here, I used DrawBot as a convenient tool. 
https://github.com/dpmellis/Font-Complexity-Data

Thomas Phinney

What is “linear density” as applied to glyphs? The proportion of the advance width that has ink, between descender and ascender? Or...?

dpmellis

It is the area of the glyph divided by its width. I don't know how meaningful it is, but it is useful for spreading out the overview graph to a second dimension.

Thomas Phinney

By “width” do you mean advance width?

dpmellis

No it is the width of the shape, so no side bearings.