[gs-devel] Re: A CJK expertise wanted

mpsuzuki at hiroshima-u.ac.jp mpsuzuki at hiroshima-u.ac.jp
Tue Feb 21 22:07:46 PST 2006


I guess you're thinking of "ideograph",
do not call CJK scripts as "hieroglyph".
http://ghostscript.com/pipermail/gs-code-review/2001-December/001549.html

On Tue, 21 Feb 2006 14:23:46 +0300
"Leonardo" <leonardo at artifex.com> wrote:
>I would like to know better about the upper horizontal arm of Su.
>In the font design space it is not exactly horizontal.

Yes.

>The question is whether we need to preserve
>the slope in raster when its size is close to the pixel size.

It's difficult to give generic conclusion.
For first, I restrict my scope to Japanese script.

I attached a few screenshots of Japanese bitmap
fonts. The design policy of glyph in bitmap font
is different from that of vector font, but I think
they gives some information about a "tendency" of
glyph design for low resolution devices.

1. Hiragana and Katakana

1-1. Comparison of font styles

The number of strokes in Hiragana and Katakana is
smaller than that of most Ideographs. Therefore,
there is room to choose flat strokes, sloped strokes,
or curves. I compare typical 2 style for Japanese
PostScript environment: Gothic and Mincho.

1-1-a. Gothic - no serif style
Especially for "Gothic" styles, the angle of stroke
is not important. See shinmk12.png, shinmk14.png,
shinmk16.png. You can find that the number of slanted
strokes are small. Especially, for horizontal strokes.
So, for "Gothic" styles, small slope for horizontal
stroke can be neglected and the normalization to flat
horizontal strokes may be acceptable.

1-1-b. Mincho - with serif style

See shinmk12min.png, shinmk14min.png, shinmk16min.png.
You can find the number of slanted stroke and curves
are greater than those of "Gothic". I think, it's to
match the stroke with the serifs at the end of strokes.
If the slope of strokes are neglected, and the serif
are attached to flat strokes, the glyphs may be ugly,
slightly.

1-2. Discussion

In the low resolution device, the bitmap font derived
from "Gothic" is easier to read, because bitmap font
derived from "Mincho" has more decoration which appear
ugly in low resolution device. For example, the default
screen font of classic MacOS, nameed "Osaka", is based
on Heisei Kaku Gothic, although standard font for printing
is Mincho (as "Ryumin-Light" on Japanese PS printer has).
>From the viewpoint of readability, when the resolution
of output device is very low, removing both of serif
and slope are expected. The requirement of slope is due
to the exist of serif, but, keeping small slope (and its
angle) itself is not important issue, I think.

2. Ideographs (Kanji)

2-1. Comparison of font styles

The most ideographs has many strokes, and less flexible
to choose flat stroke, sloped stroke and curves.

Compare shinmk12.png and shinmk12min.png,
    and shinmk14.png and shinmk14min.png,
    and shinmk16.png and shinmk16min.png.
You can find the Ideographs are completely same. Although
this example is too extreme, the difference between
"Gothic" and "Mincho" ideographs in bitmap fonts are subtle
in comparison with that of Hiragana/Katakana. In most case,
the difference between "Gothic" and "Mincho" ideographs
in bitmap fonts are exist of serif (in Mincho) or not
exist (in Gothic). The difference of slope angles, curves
are not popular.

2-2. Discussion

See 12-14-16.png and ms-12-24.png. You can find that
the increase of resolution is not used for slope or
curves. So, I think, the keeping small slope (and its
angle) itself is not important issue, I think.
Although the normalization of small slope to flat stroke
is acceptable, the changing of stroke-width looks very
ugly. This is why I prefered hint3 in previous post.

>So I would say that hint2 diminishes the slope,
>and hint3 enlarges the slope.

>The question is whether the slope is important for reading.
>If it is, we should prefer hint2.
>Otherwise we should prefer hint3,
>because it looks more smooth.

The preserve of slope is not important so much,
so hint3 is prefered?

>Also I'd like to know whether the importance is same for
>all CJK languages, or may depend on language.

I think, it depends on historical situation of printing.
In Japanese PostScript printing, "Mincho" and "Gothic"
are most important, but Chinese/Taiwanese/Korean situation
may differ. In fact, "Song Ti", a popular font style in PRC,
has remarkable contrast in stroke-width for horizontal
and vertical stroke (horizontal stroke is thin, vertical
stroke is wide). Taking a look on bitmap "Song Ti", the
number of curves in Hiragana/Katakana is smaller than
"Mincho" (see gb16st.png). I'm not sure which hinting
algorithm works better for such fonts.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk12.png
Type: application/octet-stream
Size: 26322 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk12.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk14.png
Type: application/octet-stream
Size: 21507 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk14.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk16.png
Type: application/octet-stream
Size: 22718 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk16.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk12min.png
Type: application/octet-stream
Size: 26814 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk12min.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk14min.png
Type: application/octet-stream
Size: 22306 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk14min.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: shinmk16min.png
Type: application/octet-stream
Size: 23479 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/shinmk16min.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 12-14-16.png
Type: application/octet-stream
Size: 10821 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/12-14-16.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ms-12-24.png
Type: application/octet-stream
Size: 1135 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/ms-12-24.obj
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gb16st.png
Type: application/octet-stream
Size: 12927 bytes
Desc: not available
Url : http://ghostscript.com/pipermail/gs-devel/attachments/20060222/ada028ba/gb16st.obj


More information about the gs-devel mailing list