Ellen Lupton’s ATypI Lisbon presentation on the emerging Free Font Movement mentioned Metafont in a way that brought the interpolation magic I’ve been fascinated with into sharp focus:
Metafont, created by the computer scientist Donald Knuth in the early 80s, claimed to be the ultimate realization of the universal grid. Knuth analyzed typography into a set of geometric attributes: weight, stress, width, serifs, contrast, and so forth. Manipulating these variables could, in theory, generate an infinite number of distinct yet related typefaces. It could serve, indeed, as a rational description of any latin typefaceâ€”past, present, or future. Metafont was, in short, a universal typography machine.
This sounds remarkably like my good friend Gustavo Ferreira’s Elementar system for bitmap fonts, that he presented at last year’s ATypI Helsinki.
Searching around on Metafont, I found a TugBoat article from 1998:
Knuth said that success with METAFONT would depend on “collaborative efforts between artists and programmers”. One wonders if these classes of people ever meet, because despite the earnest hopes of many, the elegant mathematical and linguistic power of METAFONT has seen little application to font design, and has not been recently used by the commercial type-design industry. Taking in hand once again those beautiful Volumes C and E of Computers and Typesetting, and observing the erudition therein, one wonders how such magnificent engines have seen such little use. Have METAFONT and Computer Modern become museum pieces, like some polished-brass steam engine, impressive to look at, but long since superceded by higher-powered technology? Or were they perhaps ahead of their time, and not yet harnessed to their potential to create? Whatever its virtues as a field of human endeavor, working with type as software seems to stifle one’s yearning to abstract and perfect a physical enterprise in mathematical form. Instead, it seems to stir the passions for raw, un-parameterized, barehanded manipulation of perimeters. You want to grab a shape like a piece of hefty rope, not tweeze some bits of code. Thus fonts today are drawn using direct-manipulation, CAA (computer-aided agony) tools that somewhat speed the brutish task of digitizing and refining outlines. Big publishers have in-house software, and the small-time designers have GUI software such as FontLab, Fontographer, and Type Designer. This author is a true believer in languages as a means to use computers. In respect of font design this could hardly be better implemented than through METAFONT. Yet when it came to the practical problem of creating a few fonts in the shortest time, even these near-absolute principles fell to the expedience of the GUI tools. It is indeed faster to just click and drag, just not to be recommended as a steady job, if you value your sanity.
The article goes on to explain in technical detail how Metafont outputs bitmap fonts from its typeface parameters, because when TeX was created the problems with outputting vector outline fonts was too hard. But for Metafont to have any relevance, this must be solved:
The key, therefore, is the ability to convert METAFONT designs to outlines.
This sounds like the kind of thing that my font wizard friends were talking about back in March, about the serious problems a Free Font Development Tool would face up to. But if FontLab can pay a bunch of Russians a euro an hour to solve them, the Free Software community can solve them too.
And if this can be done then a GUI application to set Metafont parameters interactively would be insanely great. Especially if it was coupled with Spiro foo in a bidirectional way. Like being able to code simple Scribus files, open and edit them in Scribus, then programmatically change them again, then re-edit visually. And that sounds kind of like the commandline-gui hybrid interaction Mike Cutler showed me in AutoCAD back in like 1997. All roads lead to lisp, eh…
The Metafont by David Crossland, except the quotations and unless otherwise expressly stated, is licensed under a Creative Commons Attribution-Share Alike 3.0 Unported License.