[XeTeX] can you advise me about Chinese fonts and xelatex?

Arthur Reutenauer arthur.reutenauer at normalesup.org
Wed Jan 21 07:15:54 CET 2009


> I wish the free/open fonts gave a nice result, but the students say
> neither the GNU Unicode nor the Unibit fonts (WenQuanYi Zen Hei) look
> quite right.

  GNU Unicode is a bitmap font that has been converted to TrueType
format.  It can be useful at small point sizes, like in a browser or a
text editor, but should not be used in any document meant for printing,
like a PDF file.  It simply doesn't have the typographical quality of a
text font.  This is the main reason why it looks wrong to your students:
it's designed to "look right" at small point sizes, and a text font
needs to have a different look and feel.

  In addition, the font really is bitmap (probably originally 16 pixels)
and doesn't scale at all: if you magnify it, you will see the pixels.
Incidentally, that's also the reason why it seems bolder: at smaller
point sizes, the strokes need to be a bit thicker to give the impression
of a regular font.  To sum up, it's a mistake to use this font in a PDF
file, for both aesthetic and technical reasons.

  The Wenquanyi font, while it doesn't have the latter, technical
drawback (it scales properly), simply looks ugly to me.  Some characters
seem to be completely disportionate (e.g., on the last line, the 4th
character from the end, meaning "middle", has a much to wide rectangle
for my taste).  Since I read on its Web site that it was also,
originally, a bitmap font, I suspect that this is the reason for these
seemingly ill-assorted shapes: the designs are also meant to be seen at
small point sizes (but you can at least scale them while retaining the
smooth aspect of the glyphs).

  The other fonts are OK, with the caveat that Andrew noted: the "AR PL"
fonts (AR being short for Arphic) come in two sets of characters, one
with traditional characters, the other with simplified characters.
Since you're typesetting simplified characters here, it's the latter set
you want to use, hence the ones whose full names end in "CN" (for China,
by contrast to Taiwan, I presume).  The name "Big5", in the other ones,
comes from the main character encoding for Traditional Chinese.

  There seem to be minor problems with the Arphic fonts, like the full
stop flying in the middle of the line of text instead of lying on it,
but that's really a detail that you can sort out later.  And their
license is as permissive as one can wish (it seems to be acceptable to
Debian, which should be good enough to anyone).

  As for your last question about whether the XeTeX version would make
any difference, it wouldn't.  It's all about the choice of fonts.  To
insist on it once more: avoid, at all cost, fonts that come from bitmap
fonts.  It's not quite clear to me why people would invest time into
developing huge fonts that derive from bitmap fonts today, but it's
certain that it's not a good choice for typesetting.  Also, I would
personally advise against any font that attempts to be "comprehensive",
covering all the 70000+ Chinese characters that Unicode contains today.
None of your students will ever need them (or if they do, they should be
aware of their needs and invest a lot of time into looking for the right
font :-) and this quantity will always, in my opinion, come at the price
of quality.  It's much wiser to stick to fonts that cover exactly the
character set you need, in that case, the set of simplified characters
needed to write Modern Chinese (and that's big enough!).  As Andrew
said, most "normal" Chinese fonts have a few thousand characters, not
tens of thousands.  This is exactly the case for the Arphic fonts.

	Arthur


More information about the XeTeX mailing list