Re: [Freetype] fractional point sizes

[Werner LEMBERG]

> >The point size is indeed rounded to full *pixels*, but nothing
> >prevents you from using fractional *point* sizes.  You probably mean
> >fractional pixel sizes, don't you?
> 
> Well, I'm actually a little confused about font/point/pixel sizes.
> When I talk about font size it's in points: MetaNormalLF Roman 11.5
> points.  (Apparently fractional point sizes are pretty common in the
> print industry.)

Especially TeX uses fractional point sizes all the time.

> I'm not entirely sure how the pixel size is used.  At first I was
> thinking that it was points per EM, but I'm really confused after
> reading the last paragraph in the 'note' for FT_Set_Pixel_Sizes.
> 8x8 cell??? You're definitely not talking about ppem.

Basically, FT_Set_Char_Size() and FT_Set_Pixel_Sizes() are the same.
The goal is to compute the size of the EM box in pixels -- you
certainly know that the EM box is nothing real but a means to express
certain dimensions in font-specific units (think of the `em' unit in
TeX or `m' in groff).  The remark only warns that the EM box is not
the smallest bounding box of all glyphs of a font.

The former function is better suited for outline fonts, the latter for
bitmap fonts; the used formula for a given size `s' (in points) and
resolution `r' (in dpi) is

  p = (s * r) / 72

rounded to an integer value.

> Also, I didn't think fractional pixel sizes were possible?

David and Vadim are considering this right now...


    Werner