Tuesday, August 29, 2006

Close enough for gummint work

Someone raised the issue the other day, over on the comp.fonts news group, of optical size of type, and another poster asked for an explanation. If the finer points of typography are something you’re interested in, here’s an edited version of the explanation I posted:

If you’ve looked into how digital fonts are constructed, you may think of a font in terms of a single set of outlines for some number of glyphs. If you want to set 1 pt type, you render the outlines at size 1. If you want to set 72 pt type, you scale up the same outlines to size 72. Ta da. You’re done.

However, this really doesn’t work, in terms of the way we perceive letter shapes.

If we go back in history to when artisans cut each punch by hand, we can see immediately that a character of, say, around 6 pt (this was before the invention of the point system) and a character cut by the same punchcutter of, say, around 36 pt are quite different. And the difference is not because of human error or imprecision in wielding the gravers and counterpunches. No, the difference is systematic across the font. The punchcutter’s intimate knowledge of the way type impresses ink into the surface of paper taught him to make the counters (the open spaces such as the middle of a letter o) proportionally larger in smaller sizes and to make stems proportionally thinner, etc. Visually, the range of sizes represented an integrated design, within the limits of human ability to cut consistently. However, optically magnifying the 6 pt type to the size of the 36 pt type would make them look like entirely different typefaces to the naive observer.

Jump ahead to the nineteenth century. Now we have a point system, but we’ve graduated from hand cutting each punch to mechanical punchcutting, using a pantograph. The typeface designer works with pencil on paper at a large scale (varying by designer, but often ten times the target size), then inks the drawings. A craftsman transfers the the pen sketch to a brass template, which is then screwed to a block. A pantograph traces the brass template and cuts a punch at the desired small size. Now this is an industrial process, with expensive labor and machinery. So the designer cheats the system a bit and draws perhaps three versions of the font, an A size, a B size, and a C size. The A templates are converted by the pantograph to punches in the 6 pt to 8 pt range. The B size template becomes the 9 to 12 pt punches. The C size is used for 14 to 24, perhaps. That sort of thing. The visual principle is the same as before, but there are fewer gradations. Still, if the mechanical craftspeople are particularly skilled and attuned to the problem, they can make vernier adjustments to the pantograph and other adjustments to the casting moulds so that there is a somewhat smoother gradation among sizes than you would at first suspect from having just three models.

Now we jump ahead to the mechanical typecasting machines (Monotype, Linotype, Intertype, Ludlow). These have severe limitations in terms of permissible character widths. The best that can be said is that the cost of typesetting was greatly reduced. The spirit was willing in the matrix manufacturing companies, but the technology was weak. The quality was just barely close enough for government work. We enter a dark period in design when a very few excellent craftspeople push the limits of what the machines permit, but there’s a lot of just awful type produced that mostly drowns out the good stuff. However, even in this environment, the concept of A, B, and C master designs persists, as it does into the era of phototype masters (Linofilm, for example). Meanwhile, we’ve moved from metal pressing into the surface of paper to offset technology, where a clean shape is laid on top of the paper surface. This completely changes the relationship of glyph shape (as cut) with its visual rendition (as printed). The old type designs were cut slighter than the desired impression, to allow for the three-dimensional impression and the spread of ink under pressure. The same designs, used for offset, produce a wan, weak character on the paper.

One last jump. Digital type, as outlines (there were other technologies, but nowadays we deal with outlines). What has been lost?

That’s what the thoughtful criticism of specific fonts and specific font companies is about; and, in particular, that’s what the discussion of scaling is about.

Early versions of digital fonts were an amazing improvement over what a secretary could accomplish with a Selectric typewriter. But they were a poor imitation of real commercial typography. Today, though, after the same font names have been attached to several revisions of those first outlines, and after the technical standards for font outlines have been revised a few times, and after rendering engines have gone through several generations of improvement, the basic font outlines are, in some cases, quite good.

We still have the scaling problem, though. The fonts that take this problem into account are expensive. Adobe has lately issued a few; and other font companies—I hesitate to call them foundries, as there is no metal being cast—have also addressed this issue. When you pay a professional typographer to design and produce your book, you have every right to expect that the highest quality fonts be used. These are not the fonts that came with your computer or with your version of Microsoft Office. They’re professional tools for professional use.


Nick Belardes said...

This is a good post... makes me think of younger generations working on vector-based programs and if they will even stumble across a post like this in their lifetime as they work for some hip-hop magazine drawing pimped-out rides in Illustrator, importing into InDesign, and adding cool san-serif style headlines that read, "I'm yo pimp daddy..."

Well, I care anyways... so keep the history coming...

Dick Margulis said...

Whadday mean, history? I'm old, man! I actually remember this stuff.

Susan Jones said...