Re: [ft] FT_Get_Kerning

[Graham MacDonald]

Hi Werner,
Thanks for your reply.  I'm using the latest release of freetype, but
it seems not to have been a problem after all.  The root of the
problem was my lack of understanding of fonts.  Looking at the font in
FontLab, there should have been no kerning in the pair 'íd', but the
'í' character renders outside the bounds used for fitting - a
situation I failed to consider.

So thanks for your help, but freetype is not at fault here.

--graham

On 12/12/06, Werner LEMBERG <address@hidden> wrote:
> > I've been using freetype to get the kerning pairs of a font.  [...]
>
> Which FreeType version?  I suggest to retry with the current CVS from
> FreeType.
>
> > This font was initially OpenType, but I was having trouble gettting
> > the pairs I wanted, so I converted it to TrueType - this had the
> > same problems, but FT_HAS_KERNING() returned true in this case, so I
> > kept the ttf.
>
> Which program have you used for the conversion?  Kern values can be
> either stored in the `kern' or in the `GPOS' table.  Maybe the program
> has created just a simple `kern' version, holding the complete data in
> `GPOS'.  FreeType doesn't handle GPOS at all since this has to be done
> by a higher-level library (Pango, for example).
>
> > The problem I'm seeing is that FT_Get_Kerning returns a kern value
> > of 0 for the pair í,d, where í is a latin small i with acute
> > (0x00ed).  This results in a large gap between the characters.  Word
> > and Gimp both seem to have a negative kern to reduce the gap.
>
> Well, do those programs use the OT or TT version?  AFAIK, Gimp uses
> FreeType.
>
> > First, is this the correct way of getting the kern info?  Second, if I
> > call GetKerningPairs in win32, I get 5973 pairs, even though the font
> > has 246 glyphs - I would expect more pairs.
>
> 5973 is a quite large number of kerns.  This seems to be the right
> value.
>
> In case you still have problems please send me the font privately for
> further investigations.
>
>
>     Werner