RE: IDS rendering and IDS analysis (was RE: Problems/Issues with

Date: Wed Apr 12 2000 - 05:19:00 EDT

Deborah Goldsmith wrote:
> I would think it would be really hard to get good-looking
> glyphs composing
> on the fly, but I suppose if you are just interested in
> legibility it could
> work.

I think that it is definitely possible, in theory, to render an arbitrary
non-anticipated Ideographic Description Sequence. After all, we humans can
do it just using pencil and paper :-)

An ugly, hardly-legible output could even be achieved with bitmapped fonts
and a naive rendering function.

But to reach a good typographic-level output, the only way I see is what I
call "the Milo's approach". Thomas Milo and the company he leads (Decotype,
Amsterdam, The Netherlands) have developed an incredibly sophisticated
technology for the rendering of the Arabic script. I don't know the details
(I have just seen a short presentation of it, and who knows what I
understood!), but I can describe it only with these words: a software
emulation of an 18th century ottoman calligrapher!

Similarly, a quality renderer for IDS's should try and emulate a Chinese
calligrapher and his inked brush. A "font" should, probably, just include
the basic strokes of Chinese calligraphy in all their possible variations.
The rendering engine, just like a real calligrapher, should be able to use
these basic strokes to draw each one of the squeezed components that compose
the IDS.

So, John Jenkins is absolutely right when he points out the complexity of
such a thing. The real question is "would that be worth doing?" And the
answer is, very very probably, "nope!"

But it's a pity...

_ Marco

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:01 EDT