comparison TOOLS/subfont-c/README @ 1676:1f3030b34f5c

Described encodings issues in more details.

1675:8ccac62ddfb7

and user-supplied encodings (also multibyte) through custom encoding files.

I prepared also Type 1 font `osd.pfb' for OSD characters based on bitmaps
created by chass.

Subfont was tested with Korean fonts from truetype-fonts-ko-2.0-1k.noarch.rpm
I found on http://rpmfind.net/ and euc-kr encoding.  Custom encoding file
for euc-kr was generated from charmap I found in /usr/share/i18n/charmaps/EUC-KR.gz
(glibc package).  This should work with -unicode switch for mplayer
(though it is not Unicode encoding).
It took about 10 seconds to render over 8000 characters on P3 @ 600MHz.


Custom encodings:
~~~~~~~~~~~~~~~~~
For each character you want to render write the line consisting of:
 hexadecimal Unicode character code,
 followed by whitespace,
 followed by hexadecimal number representing your encoding,
 followed by new line character.

Example: to render a single letter `aogonek' (Unicode 0x0105) and encode
it using iso-8859-2 encoding (0xB1), your custom encoding file will consist
of a sigle line:
0105 B1


New font.desc format (proposal):
~~~~~~~~~~~~~~~~~~~~~==========~
Subfont will generate new font.desc format when compiled with NEW_DESC macro defined

Notes:
~~~~~~
  + Starting x position of each character and the bitmap width is aligned
to multiple of 8 (required by mplayer).

  + Currently subfont won't work on big-endian systems.

  + My development platform is RedHat 7.1.  FreeType versions tested are
2.0.1 through 2.0.4.

  + FreeType library has a bug that makes subfont display some warning message

1676:1f3030b34f5c

and user-supplied encodings (also multibyte) through custom encoding files.

I prepared also Type 1 font `osd.pfb' for OSD characters based on bitmaps
created by chass.


Encodings:
~~~~~~~~~~
You can get any encoding and any charset.
1. If you want 8-bit charset, which is known to libc, encoded either in 8-bit
   or Unicode (like ISO-8859-*, KOI8-*):
   
   Find correct encoding name using `iconv --list' (on RedHat) and use it.
   For latin2 subtitles I would write:
   ./subfont iso-8859-2 24 verdana.ttf
   and for UTF-8 subtitles with latin2 charset:
   ./subfont --unicode iso-8859-2 24 verdana.ttf

2. If you want encoding not known to libc or non 8-bit (like EUC-KR):

   Create file describing your charset:

   For each character you want to render write the line consisting of:
       hexadecimal Unicode character code
       followed by whitespace
       followed by hexadecimal number representing your encoding
       followed by new line character
   or (for UTF-8 subtitles):
       hexadecimal Unicode character code
       followed by new line character.

   Example:
       To render a single letter `aogonek' (Unicode 0x0105) and encode
       it using iso-8859-2 encoding (0xB1), your custom encoding file will consist
       of a sigle line:
       0105 B1

       or to get unicode font.desc, write only:
       0105

   Subfont was tested with Korean fonts from truetype-fonts-ko-2.0-1k.noarch.rpm
   I found on http://rpmfind.net/ and euc-kr encoding.  Custom encoding file
   for euc-kr was generated from charmap I found in /usr/share/i18n/charmaps/EUC-KR.gz
   (glibc package).  Simple script for this you will find in encodings directory.
   This should work with -unicode switch for mplayer (though this is not Unicode).
   It took about 10 seconds to render over 8000 characters on P3 @ 600MHz.


New font.desc format (proposal):
~~~~~~~~~~~~~~~~~~~~~==========~
Subfont will generate new font.desc format when compiled with NEW_DESC macro defined

Notes:
~~~~~~
  + Starting x position of each character and the bitmap width is aligned
to multiple of 8 (required by mplayer).

  + Currently subfont won't work on big-endian systems. I need help.

  + My development platform is RedHat 7.1.  FreeType versions tested are
2.0.1 through 2.0.4.

  + FreeType library has a bug that makes subfont display some warning message