Re: [ft] Suggestion to improve auto hinter

[Collyer, Oliver, SI]

On 19 Feb 2011, at 08:20, "Werner LEMBERG" <address@hidden> wrote:

>>> This is probably dependent on the font.  For example, I've just run
>>> ftdiff on pala.ttf, and the autohint's vertical size of glyphs is
>>> basically always the same as with the unhinted glyphs.  Do you have
>>> this `problem' all the time, regardless of the selected font?
>> 
>> Yes it's dependent on the font or size or something - that's the
>> point. Some fonts don't change height, some do.
>> 
>> I think verab at a requested height of 10 pixels ended up at 9
>> pixels after light hinting.
> 
> NEVER believe that you can request a height of a glyph!  You can
> request rendering for a certain point size at a given resolution (or
> directly ask for a certain ppem value), everything else is in the lap
> of the gods :-)
> 

Is there a way to retrieve back the pixel height of a glyph, after hinting? 

Or is that something only the Gods know :)

What I mean is, if I use Set_Pixel_Sizes and let's say I ask for pixel height 
of 10. Now let's say the hinter snaps things so that the height is actually 9 
when I come to rasterise it  and view it on screen. Now is this number 9 
accessible anywhere? Or do I need to convert it to a bitmap and measure it 
myself?

Once I have a reliable way to establish what the hinter has done it should be 
easy enough to modify my application to perform some pre-analysis on a font and 
establish what point sizes are required to produce given heights.

I've hooked FreeType into a custom Windows build of Google Chrome, which I'm 
using via the Berkelium library to experiment with creating a game UI system 
and there is currently a frequent mismatch between the pixel sizes when Chrome 
runs on Windows compared to when it runs on Mac, which I want to eradicate.


> 
>    Werner