Text encoding support

Support for non-ASCII (multibyte) languages

If you have a multibyte encoded dialog.tlk you will have to add the Encoding setting to your gemrb.cfg. Set Encoding = to your language and make sure that an ini file matching your language exists under unhardcoded/shared. If no ini file matching your language exists, one must be created.

The TTF plugin must be either compiled with iconv support, or you must convert your dialog.tlk (see tlk_convert) to UTF-8 to use a non-Unicode compatible dialog.tlk/font.

To create your own language.ini, make a new text file like this (Chinese example):

TLKEncoding = GBK

Consult the following sections to learn what encoding your dialog.tlk is in. Also check out the format documentation.

After you have a working language.ini, you should be able to play using either Unicode compatible TTF fonts or the BAM fonts supplied with your language pack.

In case the original strings are in a different encoding and you want to use common fonts, you can use tlk_convert to convert them to utf-8 first.

Language overview

Here is an incomplete overview of known non-English versions of IE games. This is documentation of what we are facing to support non-ASCII characters (and other languages in general, but that’s more tricky due to IE engine limitations).


Uses CP1250 encoding (1 byte per character). The language uses cases, genders and other features making a perfect translation impossible with IE.


Polish for BG1 uses an ad-hoc encoding invented by CDProjekt. It’s definitely not any Polish encoding mentioned at Wikipedia.


One of the least problematic TLKs, requiring just a few German characters remapped for proper display.


There seem to be at least two different translations of BG1 - one has strref 15415 “фильмы”, the other has “ролики”. At least the first one uses cp1251 encoding, where cyrillic letters have codepoints 192-255, first uppercase and then lowercase letters. Unlike in our default encodings, uppercase of code 255 is code 223, not code 159.


Korean patch for PS:T uses CP949 encoding, but many strings in the TLK were left in French. Additionally some of the French strings contain some garbled characters (possibly Korean again) looking like a haphazard mix of 2byte and 1byte characters, that is not utf8 :-) )

The PS:T patch consists of

  • Dialog.tlk
  • Torment.exe


BG 1 and 2 Have two different Chinese versions, with unofficial corrections to them that can all be found at The Ring Of Wonder

The BG1 patch uses a custom executable, probably to add double byte support. It can be found at the site.

BG2 Seems to have this natively, just follow the instructions in the readme to enable.

Both patches simply replace realms.bam and normal.bam fonts and dialog.tlk - the other fonts are mapped to one of the aforementioned.

Patch for PS:T uses CP950 (Big5) encoding, that’s according to wikipedia used on Taiwan.

  • cachemos.bif
  • CREFiles.bif
  • CS_0404.bif
  • dialogF.tlk
  • Dialog.tlk
  • GTRSCRN.mos
  • Interface.bif
  • setfont.exe
  • _Torment.exe


The official BG2:SoA patch uses CP932 (Shift-JIS) encoding. It adds a possibility to switch alphabets between katakana and hiragana with F3, probably for the input. It also does some tweaks to improve legibility that might require code changes. The font is in a BAM file that contains LOTS of empty positions and in addition to latin characters and Japanese it contains Russian as well.

  • BGMain.exe
  • override/floattxt.bam
  • override/Realms.bam
  • override/Normal.bam
  • dialogF.tlk
  • dialog.tlk

(It also contains lot of files in override/, but those are bugfixes not related to Japanese)