[texhax] Request Some feed back on TeX

William F. Adams wadams at atlis.com
Thu Feb 10 16:43:17 CET 2005

I'd said:
>> I think a really promising thing to research would be to create an
>> analogue to Jonathan Kew's XeTeX which isn't tied to Apple's Mac OS X
>> platform, say coupling it with Graphite rather than Apple's text
>> object.

and On Feb 10, 2005, at 9:33 AM, Chris Rowley wrote:

> I am fairly sure that the design of XeTeX is not tied to Apple's
> technology; it is just that this is the only implememation of such a
> system so far (or is it?).  It would be better to adapt XeTeX to give 
> generic
> support for `large fonts accessable only via Unicode slot numbers'.

XeTeX is the only such implementation AFAIK. I'll grant that if the 
design / interface is really clean, it might be trivial and 
uninteresting to hook it up to an API / system analogous to Quartz and 
the AAT / ATSUI system.

> As Jonathan is well aware, it is also not a full extension of TeX to
> the world of modern fonts since it does not extend the use of `fonts
> in maths' nor does it adress the problem of glyph selection from a
> range of fonts offering different glyph collections: it assumes that
> every Unicode slot used will be sensibly rendered by any font used.
> It also works by disabling some of TeX's abilities to typeset text in
> strange ways (perhaps this should be viewed as a much needed extension
> to TeX:-).

There's been some interesting discussion of this sort of thing on the 
XeTeX mailing list.


>> I don't think hinting is going to be particularly necessary in the
>> future --- at some point, people are going to figure out that Go Corp.
>> had the right idea, make _all_ user interface elements resolution
>> independent and scalable, and _finally_ start to take full advantage 
>> of
>> high-resolution displays which will eventually become ubiquitous.
> Doh! Speaking as a mere technologist in this area, the abstraction of
> the concept of `hinting' (rather than a particular implementation) is
> central to making graphics `resolution independent and scalable'.

IME, one can get better rendering of type in medium-resolution 
greyscale / colour bitmaps by _not_ allowing hinting to kick in and 
distort the outlines by grid-fitting, especially if the type isn't 
hinted well to begin with.

Xerox has 300dpi LCDs for use in their copiers last I checked. At some 
point in time, one will get LCDs with higher resolutions and decently 
quick refresh rates, and graphics cards will improve to the point where 
they have sufficient memory and processing power and bandwidth to 
manipulate such in an interactive fashion.

Try taking a page of text-sized type in a .pdf or .eps --- rasterize it 
in PhotoShop or some other tool which groks hinting at a resolution 
which allows hinting to kick in, but which is still high enough to 
promise decent representation. Take the same page, convert all the type 
to paths, convert it to a bitmap at say twice that resolution, then 
downsample it to match the first sample.

I've found that the second process yields a result more in keeping with 
a printed page with better, more even spacing, and more legible in most 
instances w/ most typefaces.

>> Anti-aliasing ameliorates the need for hinting by a great deal as well
> That sounds counter-intuitive and, anayway, why does it need 
> ameliorating?
> Actally `ameliorating the need' increases that need, if it is a
> `good need' in the first place (as I claim):-).

``Anti-aliasing ameliorates the unpleasant situation of needing hinting 
by a great deal.''

better? ;)



William Adams, publishing specialist
voice - 717-731-6707 | Fax - 717-731-6708

More information about the texhax mailing list