[MPEG-OTSPEC] Some research on TT instructions and cubics

[Georg Seifert]

> Am 22.09.2023 um 22:52 schrieb Laurence Penney <lorp at lorp.org>:
> 
> On 22 Sep 2023, at 20:28, Skef Iterum <skef at skef.org> wrote:
>> Setting something like this up and trying some manual hinting of a few glyphs
>> might be the quickest route to confidence about how cubics will interact with
>> TT hinting. (Autohinting will presumably take a while longer to sort out.)
> 
> One method that might just work is to experiment on fonts with quadratic sources made via the "FogQ method". This is a restriction built into the design tool (Robofont these days) such that curves always use two consecutive off-curve points between on-curve points, leading to a pretty familiar editing experience for designers used to cubics. Such fonts should convert well to cubic fonts where each cubic glyph has exactly the same point structure as the source quadratic glyph, with the two off-curve points per curve repositioned automatically. And, because a) human- and auto-hinters almost always touch only on-curve points, and b) these days mostly avoid delta hints, an existing set of quadratic hints (the very same byte strings!) will likely behave well. The set of fonts you could try includes Amstelvar, Roboto Flex and most fonts that David Berlow & crew have worked on in the last few years.

Quadratic curve have some disadvantage over cubic as they are less versatile on what curves they can represent and often need more segments. But the trick with the implicit oncurve point between consecutive offcurves allow certain things. E.g. there are no kinks. So limiting the designer to only two offcurve points it a really bad idea. 

georg