Home / ATypI Tech Talks 2021

Parametric Type Classification in the Context of Letter Fitting

“The topic of classification emerges in a variety of type design contexts. Historians tracing the evolution of typefaces or of individual letterforms, educators framing stylistic questions for designers, and vendors superimposing structures on vast type libraries are all classifying type, but they are performing distinct tasks. In his own research, which explores algorithmic letter fitting in type design, Nathan Willis has encountered the need to classify typefaces and letterforms in yet another situation. Successful letter fitting depends on the combined impact of numerous design characteristics: weight distribution, width and proportion, contrast, slant, and so on. To adequately assess a fitting algorithm across such a large design space, Willis needed to construct a set of objective measurements and quantitative metrics with which to classify the typefaces under investigation. This presentation will outline Willis’ process for devising classification metrics useful for quantifying, plotting, and clustering typefaces, including a discussion of how the letter-fitting subject poses questions that are distinct from “”weight”” and “”width”” metrics as exposed in CSS and related frameworks, and how the evaluation of practical typefaces differs from the design-time parametric approaches formulated by Type Network and Donald Knuth. Although this project is specific to letter fitting, the underlying need to classify typefaces so that they, in turn, can serve as research material, has implications for many other lines of inquiry in type research.”
“The topic of classification emerges in a variety of type design contexts. Historians tracing the evolution of typefaces or of individual letterforms, educators framing stylistic questions for designers, and vendors superimposing structures on vast type libraries are all classifying type, but they are performing distinct tasks. In his own research, which explores algorithmic letter fitting in type design, Nathan Willis has encountered the need to classify typefaces and letterforms in yet another situation. Successful letter fitting depends on the combined impact of numerous design characteristics: weight distribution, width and proportion, contrast, slant, and so on. To adequately assess a fitting algorithm across such a large design space, Willis needed to construct a set of objective measurements and quantitative metrics with which to classify the typefaces under investigation. This presentation will outline Willis’ process for devising classification metrics useful for quantifying, plotting, and clustering typefaces, including a discussion of how the letter-fitting subject poses questions that are distinct from “”weight”” and “”width”” metrics as exposed in CSS and related frameworks, and how the evaluation of practical typefaces differs from the design-time parametric approaches formulated by Type Network and Donald Knuth. Although this project is specific to letter fitting, the underlying need to classify typefaces so that they, in turn, can serve as research material, has implications for many other lines of inquiry in type research.”

`
Speaker

Nathan Willis

PhD candidate University of Reading

Nathan Willis is a PhD candidate in the Typography and Graphic Communication department at the University of Reading, UK.

Speaker

Thomas Phinney

Font detective, expert witness, type designer, consultant, writer Font Detective LLC

Thomas Phinney is a type designer and former CEO of FontLab. In the 2000s, Phinney helped drive the adoption of OpenType within Adobe and in the marketplace. Phinney developed character set standards for Adobe, and designed the Latin, Greek, and Cyrillic typefaces hypatia sans, cristoforo and science gothic. Phinney has four patents and a master’s degree in printing and typography from the Rochester Institute of Technology, where he completed his thesis on technical challenges in multiple master font technology. Phinney has been on the board of ATypI since 2004. Blog: www.thomasphinney.com