freetype-devel
[Top][All Lists]
Advanced

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

Re: Fw: [Devel] Re: The guts of yudit-2.xx


From: Antoine Leca
Subject: Re: Fw: [Devel] Re: The guts of yudit-2.xx
Date: Thu, 01 Feb 2001 17:31:08 +0100

Werner LEMBERG wrote:
> 
> Can someone please verify this claim?

I shall check Freetype 1 this evening. Currently it does not seem to
do what Alex calls for, but I did not check if this is or not a problem
and what it may affect.

> De: Alexander Gelfenbain <address@hidden>
> References: <address@hidden> <address@hidden>
 <address@hidden> <address@hidden>
> 
> If I am not mistaken load_truetype_glyf() does not do all the calculations
> prescribed by the manual.
> 
> Check out: http://fonts.apple.com/TTRefMan/RM06/Chap6glyf.html

Well, AFAIR we use the MS specs in this area.
And <URL:http://www.microsoft.com/typography/otspec/glyf.htm> has no sign
of this specific stuff.

I do not remember seeing this stuff about m and n in the past, even on Apple
online specs (I did never check the written specs, but I believe David did).

However, it is well known that Apple and MS engines differ abit on their
handling on composite. IIRC, the main known difference is that the
scaling does not take place at the same moment (this is important with
2+ levels).

Also, David (in 1.x gload.pas/ttgload.c, I did not check v.2) disable
hinting if the discriminant of the matrix is different from 1, that
is when there is scaling or slanting of the component.

A point worth noting is that Microsoft expressly asks their developpers
to stay away from using the most complex form with 4 coefficients:
<URL:http://www.microsoft.com/typography/developers/delivery/hinting.htm>
   3.2.1 Composite glyphs 
   [...]
   The composite instruction SOFFSET for performing flipping, scaling or
   other transforms should not be used.

[ Antoine's note : SOFFSET is the instruction in RoyalType which
 translates to a 2x2 matrix for a composite. ]
 
> But unfortunately this does not seem to be the cause of the Aring problem,
> at least after I modified the Freetype code to calculate m and n
> properly Aring still did not look right. Anyway I am still looking into
> the problem and I will keep you posted.

What for a font are you using? Is it some Apple font?


Antoine



reply via email to

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