RE: [Algorithms] Next generation GUI.

[Jamie F. <ja...@qu...>]

well, this is just a case of picking an appropriate level for your interface
abstraction, just like with meshes et al.

at a low level, we render each character as a separate quad. but at the
public API level, strings are passed in whole, so the positioning, etc., of
each character does depend on adjacent characters, and we do get all the
benefits of a good font, etc.

there's no reason that how you render your characters should be linked to
the formatting of the strings they're in.

jamie


-----Original Message-----
From: Jon Watte [mailto:hp...@mi...]
Sent: 13 September 2002 17:58
To: Jamie Fowlston; Jonathan Perret;
gda...@li...
Subject: RE: [Algorithms] Next generation GUI.



> we render font to texture, use quad per character. platform
> independent, we
> have it working on linux, win32, freebsd, xbox, ps2....

I guess my point is more that you get better results when you
render whole words (or the whole actual text) rather than
rendering just individual characters. The reason for this is
that (good) fonts and font renderers can (and do) make
decisions about things like kerning and ligatures, which you
cannot do when you just render a single character. (There's
also the esoteric phenomenon of stochastic fonts, if you're
way into type-setting land)

If your font is Courier New or Lucida Typewriter, then there
probably won't be a difference, though :-)

Whether you render with GDI, FreeType, QuickDraw or something
else is probably more a matter of what your target platform(s)
is/are. As you say, some libraries are more portable than
others.

Cheers,

				/ h+

				Did pre-press in a previous life