CAB@SAIL.STANFORD.EDU (Chuck Bigelow) (08/13/86)
Some time ago (May 1) Bob Morris sent out a discussion of the relationship of Metafont (a language for describing typefaces) to Lucida (a typeface) to PostScript (a language for describing graphic images, including typefaces). There was little response to it, perhaps because the subject is complicated, or perhaps because no one is interested. Hoping the case is the former rather than the latter, here is a response. For now, I want to leave out Lucida, a subject which I have discussed elsewhere, and concentrate on Metafont. To warm up, an informational notice. Donald Knuth's magnum opuses (magna opera, maybe) COMPUTERS & TYPESETTING, vols. A - E have now been published by Addison Wesley. A. The TeXbook B. TeX: The Program C. The METAFONTbook D. METAFONT: The Program E. Computer Modern Typefaces For those looking for intellectual challenges in typography, these are the items. Bob Morris's message is relevant to vols. C, D, and E. If you want to learn how Metafont works, read C. If you want to see how Knuth did it, read D. If you want to see what he did with it, read E. All these books are unique, but E deserves special attention. After seven years of work, Computer Modern has never looked better. The letters are well-shaped and the whole is well composed and printed with that crispness that the admirers of Modern love and the detractors deplore. I believe that connoisseurs of the Modern style will agree that Computer Modern is now the equal of its inspiration, Monotype Modern 8A, though it is not identical. Unlike Modern 8A, however, Computer Modern is a "metafont" in the Knuthian sense, and thus the changing of parameters can radically, or trivially, alter the design. As Knuth says, there is an infinite number of types possible from the single set of Metafont programs, but this infinite typeface universe is still bounded by "computer modernness", and is obviously only one of the transfinite typefaces possible with Metafont. Perhaps it should be called Computer Modern aleph-sub-null. The way in which Knuth has programmed the letters is the next important matter. They are neither simple outlines, nor simple vectors; rather, they are composed of complex draws with nibs often composed of a triplet of points arrayed in a line. The center point determines the skeleton of the letter and the "outrigger" points (my term, not Knuth's) can be varied to give the stroke various shapings. There are many otherr methods also used. The drawing pens do not move in an absolute grid, but in an algebraic grid, where the position of each point is usually determined by the positions of other points and the values of certain parameters. This gives the programs their protean power. It also makes them very difficult to understand. Knuth recommends that the dedicated reader actually try the programs out with Metafont, and I agree that this is probably the only way to grasp what is going on in most of the more complicated letters. I find many, if not most, of the descriptions utterly bewildering on a first (and second) reading. The interesting problem that Bob Morris raised is, how does Metafont relate to PostScript? This can be extended to the question: How do Metafont descriptions in general, and Computer Modern programs in particular (as the best documented Metafont characters) relate to outline font descriptions in general, and in particular those that use `structuring' (to avoid danger words like `intelligence' and `smarts') such as the proprietary structured outline fonts made by Adobe for PostScript, by Imagen for DDL, and by other firms without PDLs or DDLs? (Sorry about the prolix question.) The superficial answer is that the Metafonts are orthogonal to the structured PDL and DDL fonts; that there is really no relation. A more careful analysis would reveal, I believe, that Metafont and the Computer Modern types constitute the most ingenious and thoroughgoing attempt to make explicit the semantics and syntax necessary to structure a font in any graphic description language. The principles of Metafont could be extended to a PDL, or have to be in some way re-invented in a PDL, to structure a font to produce aesthetic scaling and rasterization. This is not to denigrate the PDLs or the quality of their fonts and font description methods, it is only to ask whether there is equivalence at the deeper structural level of font descriptions. It is impossible to publicly test such hypotheses, however, because the structures of PostScript and DDL fonts are proprietary trade secrets. (Note: the structures are secret, not the Bezier or conic outline formats.) Compugraphic has applied for a patent for its "Intellifont" process, however, so this may eventually become public knowledge, and if CG sues the producer of some other structured font system for patent violation, as III recently did to CG for outline font patent violation, then those details may eventually be revealed in court as well. Another point is that the Computer Modern fonts are simply one way to use Metafont. The Euler fonts are an example of another way. There is font programming "style". So far, there have been no academic discussions of good fonting style, so it is difficult to compare Knuth's style to some standard. We are still in the pioneering era rather than the bureaucratic era. It is my own opinion, possibly not shared by Knuth, that Metafont will only become truly useful for lettering designers when an interactive user interface is developed for it that hides most of the symbolic details of program construction and linear algebra from the designer, replacing it with a direct manipulation of graphical objects metaphoric display. Such an interface must not eliminate the possibilities of structuring the fonts. Not an easy task, in my estimation. Take note, o ye who are looking for something difficult to work on for a few years. How does one go about displaying a multi-dimensional parametric space, in which the number and "direction" of the dimensions must be determined by the user? It is noteworthy that Knuth has placed all this information in the public domain or in print. If you want to read the code, there are the books. If you want to try the systems, you can get the tapes. If you want to fool around with the fonts, you are welcome to have a go. This is a true contribution to typographic understanding, because it is not bounded, but becomes a foundation on which others can build. Knuth has solved the font copyright problem by giving out the fonts. A fine gesture. (No PDL based fonts offer anything like the plasticity of Metafont. Of course, most PDL based fonts are intended to reproduce existing fonts, not provide a malleable medium of font experimentation. Indeed, most typeface owners strongly resist the modification of their designs by other hands, and encrypt fonts to prevent piracy. A reasonable attitude and a prudent precaution, given the unethical practices of many font pirates.) DISCLAIMER: In volume E, Donald E. Knuth thanks me (among many) for contributing to his work on Computer Modern, and I donated a brief account of the Modern style for the book. So I may not be an impartial reviewer. However, my contributions were more often critical than not, and so I feel I have earned the right, through several years of harsh commentary, to dwell on the undeniable successes of Knuth's achievements without being guilty of puffery.