RE: FW: Type1 fonts

[Jeff Henrikson]

> I'm not an expert in what hinting is, but at least for the music
> typesetting, I don't think it would make any difference, since
> Lilypond determines the exact position of each symbol. It might

No, positioning is not (under normal circumstances) the point of hinting.  
Hinting has to do with rasterizing symbols in small
numbers of pixels (eg, on the screen) so that the important features are not 
ambiguated.  (Eg, a natural and sharp looks similar if
you don't have too many pixels.  An accent or upbow turns into a horizontal or 
vertical line, etc.)  Small numbers of pixels are of
prime importance in music notation for large instrumentation.

The reasoning sub-pixel positioning is not handled is that the font caches 
often quantize the offset of the characters to an
integer pixel, so that rasterizations can be reused.  The quantization can be 
made finer than pixels, but without some
quantization, font rasterizing time becomes prohibitive.  (Very low 
quantization may be the reason Acrobat is so slow.  But IMHO
that approach makes for nasty renderings because horizontal and vertical 
boundaries are often blurred over two pixels.  That's
especially annoying for things like staff lines.)

Hinting is a very laborious process, far more so than just the outline 
specification.  This among other things is what sets apart
fonts like Adobe Sonata from the others.  Of course, today with high resolution 
printers it only matters on screen, which is not
the focus of lilypond because it's not capable of interactive processing.

The reason why I ask about nonlinear scaling is that I don't know of any 
nonlinear scaling fonts (of the TeX legacy) which have
been hinted.  Probably because Metafont doesn't have a notion of hinting, but 
also because it would be a serious mind tweak to have
to do so.  Just for reference, high quality fonts like the Times New Roman in 
Windows are often rendered at important small points
sizes by hand on a monochrome grid.  Then hints are added until the automatic 
renderings conform exactly.  I'm not sure what they
do now that every mainstream font rasterizer supports antialiasing.


Jeff



> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden Behalf Of Mats Bengtsson
> Sent: Wednesday, July 11, 2001 12:59 PM
> To: Jeff Henrikson
> Cc: gnu-music-discuss
> Subject: Re: FW: Type1 fonts
>
>
> > FYI: the free program "pfaedit" has a type 3 parser.  I haven't tried 
> > plugging Feta into it.
> >
> > > The problem is
> > > rather difficult, btw.
> >
> > The general problem is, of course, but have you bothered to read the feta 
> > type 3 code?  There's nothing sophisticated
> in it and I'm
> > sure an experienced postscript hacker could cook up (at most) a simple 
> > recursive decent parser to do it, maybe with a
> little duct
> > tape.
> >
> > Any such conversions will still be primitive compared to a "real" type 1 
> > font for screen output because of lack of
> hinting.  Once
> > there was an outline conversion, hints could be added by an expert by hand, 
> > but that works against the "everything should have
> > source code" mentality.  IMHO not a huge deal.  And hints would be a 
> > definite improvement for a few symbols (accidentals come
> > immediatly to mind.)
>
> I'm not an expert in what hinting is, but at least for the music
> typesetting, I don't think it would make any difference, since
> Lilypond determines the exact position of each symbol. It might
> make a difference if the symbols are used within ordinary text,
> where TeX handles the positioning.
>
> > BTW- I was wondering the other day how much nonlinear scaling behavior 
> > there is in feta.  (Ie, the thing that Computer
> Modern does
> > a lot but TrueType, Type 1, fonts can't.)  I don't know enough metafont to 
> > see for myself.
>
> You mean that the proportions are different for different font sizes?
> At least for the few symbols I've designed for the Feta font, this
> hasn't been done in any systematic way.
>
>    /Mats
>
>
>
> _______________________________________________
> Gnu-music-discuss mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/gnu-music-discuss
>