C.2. Type 3 Fonts

We shall first describe Type 3 fonts, as they are the closest to the standard PostScript language that we have just described. In fact, the glyphs in these fonts are defined by arbitrary PostScript operators: everything is permitted, including drawing, filling, including bitmaps, and even typesetting with other fonts. In theory, any PostScript page could become a single glyph in a Type 3 font.^[C-9] But then why, despite this immense power, have Type 3 fonts been an unprecedented commercial flop? Why do we find hardly any at all in the world, be they free or commercial?

^[C-9] The reader will recognize here the principle behind the SVG fonts and, in part, that of the virtual fonts of T_EX.

Simply because we must not confuse the typesetting of text with graphical design. In typesetting, we use the graphical elements of fonts, i.e., their glyphs. These are characterized by three properties: their high degree of repetition (which calls for the optimization of resources), their small size, and the uniformity of their strokes. The first property (repetition) is managed very well by all types of fonts, thanks to the structure of the font dictionary. The last two properties, on the other hand, present a significant problem. If most of the capital letters have predominantly vertical strokes, the aesthetic tradition would have these strokes be of the same thickness within a single font. The printer must therefore know that it must assign the same number of pixels to each of these strokes, whatever their position or the orientation of the glyphs on the page may be. Worse yet, if we change sizes, the number of pixels in a "generic" vertical stroke will also change, but the condition of uniformity remains the same: within a single font, the vertical strokes must have the same thickness, period!