freetype-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ft-devel] ttfautohint: progress and problems


From: Werner LEMBERG
Subject: [ft-devel] ttfautohint: progress and problems
Date: Wed, 25 May 2011 07:09:49 +0200 (CEST)

The last days there was a lot of progress in ttfautohint.  I thought
that the whole framework was finished, and I only had to fill up
missing bytecode.  However, while doing so, I found out that I've made
a fundamental design error.

The idea was (and still is) to compute and collect `hinting sets' for
a large range of sizes, say, 8<=ppem<=1000.  For each size a set of
rules gets defined and compared to the previous ones.  Only different
sets get stored.  Example: glyph 1327 in `pala.ttf' (the `CE' sign)
has 9 sets for the following ppem values:

        8 <= set1 <   10
       10 <= set2 <   23
       23 <= set3 <   25
       25 <= set4 <   35 
       35 <= set5 <   44
       44 <= set6 <   48
       48 <= set7 <  168
      168 <= set8 <  503
      503 <= set9 < 1000

In most cases, the number of sets is much less (usually three or
four).

The problem is that the way I've collected this data doesn't work.  My
assumption was that it is sufficient to simply examine the edges after
the hinting has been done, then emitting bytecode for computing the
actual edge positions.  However, the autohinter does some actions
which escape this scheme, making it necessary to record the hinting
steps while they are executed so that they can be replayed later on.

In other words, I have to redo a large part of the framework to fix
this.  Fortunately, it's not too difficult, but I think it will take a
week before I can continue with bytecode stuff.


    Werner



reply via email to

[Prev in Thread] Current Thread [Next in Thread]