Re: [ft-devel] merged CFF and Type1 hinting code now in master

[Dave Arnold]

Hi Werner & Nikolaus,

I should back off and leave this a theoretical argument. The synthetic hint example I gave doesn't really fit as well as I was thinking (and may not be as useful for your purpose).

Synthetic hints are based on the concept of Ideographic Character Face (ICF). The code consults blue zones as part of a heuristic to determine the ICF, but it does not directly use the zone values to calculate the hint position. The synthetic hints affect the hintmap calculation and interact with the  charstring hints, if any, to improve "optical alignment" in CJK text.

I did find some internal Adobe documents from 2010 that describe synthetic hints in more detail, if you are interested. But, as I said, this may not be useful in the context of the current autohinting discussion.

Thanks.

-Dave

On 11/20/2017 9:15 AM, Werner LEMBERG wrote:

Hello Dave!

I consider a font broken if it has a blue zone table but no hints
– shall we really take care of that?  We don't do any harm; it's
just that the font stays unhinted.

Hm, well, yeah, it's okay I guess. Might be solved by teaching the
Adobe hinter to cope with blue zones alone in the distant future.

I repectfully disagree with Werner's statement about these broken
fonts.

Fair point :-) `Broken' is too harsh a word – what I actually meant
was that such a font can't be (currently) hinted by the new CFF engine
in FreeType.

Blue zones are simply a declaration of font properties that the
designer specifies.  It is up to the hint interpreter to decide how
to use that data.  For example, there is a heuristic in the Adobe
CFF interpreter that synthesizes charstring hints from blue zone
data.

Very interesting!  This is exactly what Nikolaus proposes, too.  Can
you provide some details?


    Werner