The black art of hinting type

In his online series of articles, type expert Beat Stamm* gives a wonderfully detailed introduction to the black art of “hinting” type, the arcane algebraic process that makes type on a computer screen look more like type, and less like children’s blocks. Warning: the article is not for the innumerate nor for the faint-hearted. It helps to have a degree in higher mathematics, which is probably why I found it entrancing and baffling at the same time: I failed General Pure Mathematics I at the University of Sydney in 1961. Boo hoo. Here it is: «The Raster Tragedy at Low-Resolution Revisited: Opportunities and Challenges beyond “Delta-Hinting”»
ArialSlash1x2x4x8xPixelCoverage
Here is the Introduction:

Rendering fonts on digital devices harbors the potential for incredible flexibility and convenience. Gone is the burden of carving individual letterforms out of heavy metals that are hazardous to the environment. At the push of a mouse button, the highly creative oeuvre of a seasoned type designer can be rendered at any type size and device resolution and on any digital output device—just because it’s digital.

Or so it seems. What tends to get lost in this MP3 generation of irrational digital exuberance is that font design, like playing a violin, is an analog process[1]. Fonts first have to be converted to digital. Now, modern recording devices can choose to convert analog to digital at an incredibly high rate of precision. By contrast, fonts have to be converted to very few pixels. There is no choice. The rate of precision is dictated by the rate of pixels on today’s screens.

Around 1990 the first scalable font formats appeared on the market trying to address this limitation. At the time, fonts were rendered mostly in “black-and-white.” Rendering seemed simple: Pixels were either “on” or “off,” for better or for worse. Fonts had to be “super-hinted” or “delta-hinted” to make them look “nice.” Since then, various “font smoothing” methods have been introduced, promising progressively “nicer” fonts thanks to “gray-scaling” or “anti-aliasing” and eventually “ClearType,” “CoolType,” or “Quartz.”

All these methods offer opportunities at making fonts look “nicer” on screen. At the same time they come with their own sets of challenges. Some opportunities or challenges are closely related, some are different, and some are mutually exclusive. One method may be better than another method in some aspects but not in others, and vice-versa. On today’s screens, I won’t qualify any method as “the single best on-screen font rendering method.” For sufficiently small type sizes on low-resolution screens, every method represents a compromise of some sort.

This website illustrates various methods for rendering fonts on low-resolution screens, along with their opportunities and challenges. It is not a “manual” or a “textbook” on how to “hint” in software ‘x’ for method ‘y’ on device ‘z.’ Rather, it illuminates various aspects of font rendering in different contexts. It is not all-encompassing, but hopefully comprehensive enough to show what the compromises are, which ones are avoidable, and which ones aren’t. With this insight, and with the implementation proposed in 6.3, understanding font rendering on low-resolution screens hopefully becomes choosing your own compromises.

© Copyright 2009-2012 by Beat Stamm. All Rights Reserved.