Re: [ft-devel] State of autohinter stem darkening

[Nikolaus Waxweiler]

> Would it be possible to implement TrueType stem darkening without
> using hinting like in the CFF case? I really think that darkening and
> hinting should be orthogonal.

I think so, too. Currently, stem darkening is basically a proof of 
concept bolted on to the autohinter. My idea is to turn stem darkening 
into a kind of generic service within FT so that all engines can 
implement it independently of the autohinter, just like the CFF driver 
does it.

A problem is getting at the standard widths of a font. The emboldening 
code needs to know how thick or thin a glyph is to know how much it 
should be emboldened. The Type 1 and CFF engines can access the PS 
private dict and get the designer's values from there and the autohinter 
simply computes them by itself (sometimes too thin for special symbols, 
leading to excessive emboldening). But there is no such shortcut for 
TrueType fonts without the assistance of the autohinter. Maybe the 
heuristic 50/75/110 values suffice for a first implementation. Or maybe 
I can change the autohinter a bit to just make it spit out those 
standard widths. I'll see. Just don't hold your breath ;)