search

FontBureau / Opentype-1.8-Axis-Proposal

Apache License 2.0
0 stars 1 forks source link

Should parametric be made relative? #69

Open davelab6 opened 3 years ago

davelab6 commented 3 years ago

Looking at @lorp 's https://lorp.github.io/parasync/parasync.html I see @tphinney has an YOPQ axis that is ranged relatively -starting at 0 and going up in per mille units.

Per recent discussion of grade, i also wonder if it makes better sense in a ot1.8 world to have that be the standard way to set up parametric axes.

dberlow commented 3 years ago

Our preference at the font bureau, in developing typography, is to use only values per em, or only OS2 values, and not to mix them.

If required by program, a bridge, or bridges can be established between per em and os/2 for calculations. But, Per mil values are somewhat useless to me and programs if they are relative to the default, no?

davelab6 commented 3 years ago

This isn't related to os/2 values per se. It's purely about per-mille values. But the mixing is in the specification of typography, yes, as users will proceed to set up the broad strokes of their typography with the former and then finesse with the latter - unless you are proposing to distribute 2 fonts, with OS/2 axes only, and with Parametric Axes only, but I'm afraid that will doom the latter to obscurity and zero billion adoption.

When parametric axes values are coupled to the default, they become useless once you move away from the default using user axes - optical size, weight, width, grade, and expressive axes - and that's why in typetools we have to bridge them. Given how many times that bridge has fallen down in the past 4 years after it was raised, I'm skeptical as to the wisdom of this approach.

Finally, it seems unlikely to me that the masters drawn for the user axes will be 100.00% replicable with Parametric Axes. I'd like to see some demonstrations proving this is the case, if so.