This reminds me of Font Chameleon back in the nineties. Although accomplished completely differently, the results seem similar.
The idea of blending between different fonts seems like a great idea, but the results you get from blending fonts, while you might get lucky sometimes, are rarely good or useful.
I think the whole concept makes some weird assumptions about how typeface design works, for instance that the difference between any two corresponding characters from two typefaces applies in a coherent way to all characters in the two fonts. But just because one character in a blended font looks good or acceptable, there’s no reason to expect that to be true for all characters.
It’s interesting, but I’m not quite sure what real problem is being solved.
Variable fonts are fundamentally different in that the entire blending space is controlled by the type designer.
Also, they claim to have taken their set of fonts from a Windows install after removing foreign language and symbol fonts. In fact, they removed only those foreign language fonts which lacked a roman subset. Since a significant number of the non-roman fonts in windows use either Arial or TNR (or Nimbus Serif) for their roman range, those fonts are going to be rather overrepresented. And of course many CJK fonts have roman sets with metrics and widths which really don’t correspond to normal roman aesthetics.
I played with Font Chameleon. It was fun, and its approach to blending worked better than some. I liked its approach to abstracting a design as a “font descriptor.”
I don’t think many people ever got to use the actual FontChameleon editor used to create new descriptors—the descriptors being the fonts you could put into the blender, and blend between multiple descriptors, or adjust their weight, width, x-height, etc.
I tried the editor, as Adobe had acquired the rights to the tech (used for font compression in printer ROMs for PostScript 3), and thought it had potential as a way to make remarkably quick prototypes of fonts, that could then be worked on afterwards with traditional tools. Some of my Adobe colleagues (at least David Lemon and Robert Slimbach) disagreed, feeling that it blandified the fonts a bit, and perhaps discouraged innovation in details.
Of course, other tools have gone down that path since. But FontChameleon was one of the first, after MetaFont. (And massively easier to use and more approachable than MetaFont.)
I had the starter version of Font Chameleon (I think) which included a small set of fonts to play with. My opinions are based on that. (Grant Hutchinson has it in his collection now, I believe.)
FWIW, Font Chameleon could do way more than the research project that is the topic of this discussion. It had sliders to control things like x-height, width, weight, and so on, in addition to being able to blend fonts.
I'm not really familiar with Prototypo, but one obvious difference is that, with Font Chameleon, you couldn't create a new font in it. You had to use existing fonts supplied by the developer.
Comments
Some copyright issues.
The idea of blending between different fonts seems like a great idea, but the results you get from blending fonts, while you might get lucky sometimes, are rarely good or useful.
I think the whole concept makes some weird assumptions about how typeface design works, for instance that the difference between any two corresponding characters from two typefaces applies in a coherent way to all characters in the two fonts. But just because one character in a blended font looks good or acceptable, there’s no reason to expect that to be true for all characters.
It’s interesting, but I’m not quite sure what real problem is being solved.
Variable fonts are fundamentally different in that the entire blending space is controlled by the type designer.
I don’t think many people ever got to use the actual FontChameleon editor used to create new descriptors—the descriptors being the fonts you could put into the blender, and blend between multiple descriptors, or adjust their weight, width, x-height, etc.
I tried the editor, as Adobe had acquired the rights to the tech (used for font compression in printer ROMs for PostScript 3), and thought it had potential as a way to make remarkably quick prototypes of fonts, that could then be worked on afterwards with traditional tools. Some of my Adobe colleagues (at least David Lemon and Robert Slimbach) disagreed, feeling that it blandified the fonts a bit, and perhaps discouraged innovation in details.
Of course, other tools have gone down that path since. But FontChameleon was one of the first, after MetaFont. (And massively easier to use and more approachable than MetaFont.)
FWIW, Font Chameleon could do way more than the research project that is the topic of this discussion. It had sliders to control things like x-height, width, weight, and so on, in addition to being able to blend fonts.