[ Contents | 3. | 5. ]

4. AP Font Library: The Current Status

Keisuke Kamimura
Center for Global Communications (GLOCOM)

Overview

The AP Font Library is a set of computer fonts for languages in the Asia Pacific region. It is based upon the AA Institute Fonts, which was developed by the Institute for the Study of Languages and Cultures of Asia and Africa (AA Institute), Tokyo University for Foreign Studies.

In October 1999, the Center for Global Communications (GLOCOM), International University of Japan, was given permission from the AA institute to make non-profit use of their computer font sets, and since then is undertaking effort to make the font sets available in an outlined format. The languages in the AA Institute Fonts includes Arabic, Traditional Mongol, Thai, Khmer, Devanagari, Bengali, Tibetan, Laotian, Burma-Mhong, Tamil, Kannada, Malayalam, and Telugu. Technical details of AA Font was given in our last year's report.

Of the font sets that were transferred to us, Khmer was chosen to be the target of work in 2000, because we had close relationship with experts in the development of Cambodian character fonts, who also are involved in the standardisation of character code. The snapshot of our current work is available as a TrueType font at the following url:

Our work is still underway and at this stage of development, the font is not suitable for a practical purpose in any way, but we will keep refurbishing it in the future.

How conversion works

As described in the 1999 report, AA Institute Font was designed for the use of main frame computers, and it needs conversion in order to be used on a PC platform. We started our conversion efforts with a group of Unix/Linux tools that are freely available. However, those tools work best with glyph data that has finer resolution than the AA Fonts, and it turned out that the outlined glyphs resulted from conversion needed far more refinement than was initially expected, which made it practically impossible to refine each one of the glyph in the font.

In 2000, we made use of the following two commercially available products.

ScanFont automatically recognises the contour of an glyph shape and generates an outlined glyph. FontLab helps to modify the shape of outlined glyphs that are created by ScanFont or by other means. FontLab also add additional information for the font.

Problems to be addressed

The following issues arose during the conversion process.
  1. aesthetics of converted data
  2. supplementation of additional information
  3. assignment of glyphs in the code table
The first one had been anticipated even before we started our work. Although each glyph in AA Font is represented in a 96 by 96 matrix, the actual shape of the glyph does not fill the entire matrix. Many of glyphs contain large margin areas, top and bottom, and right and left, so that the actual shape takes only a small portion of the matrix. Because the original data does not contain enough information to reproduce outlined data, conversion ended up with distorted contours and broken lines in many cases.

The second problem concerns how we supplement additional information to the resulted font after conversion. AA Font has additional information such as,

The image below shows how the logical width and horizontal position are represented.
10610.png
These items of additional information could be translated to our outlined font in a technical sense, but they have not been reflected yet. It takes a large amount of  'manual labour'. This remains one of the work items for year 2001.

Assignment of glyphs to the code table raised another problem. Even as of January 2001, there is no national standard for the coded character set for Khmer. We assigned our glyphs to the Limon font, one of the most widely accepted fonts in Cambodia. Limon sorts glyphs according to a proprietary character code table. It is reported that standardisation work is underway and we expect that we transplant our characters to the forth coming national starndard in the near future.

Even with Limon, there are a couple of issues to be considered. There exist glyphs that are contained in the AA Fonts, and not in Limon, and ones that are not contained in the AA Fonts, but contained in Limon. Most significant glyphs in AA Font correspond one-to-one to their counterparts in Limon, but diacritics and precomposed glyphs often do not. We simply excluded ones that do not match one-to-one this year.

Achievement in 2000

At the beginning of this year's work, we assumed that our primary work was to extract outline data from the original bitmap and to apply aesthetic adjustment. After a year long work, however, we recognised that additional information, such as logical width and horizontal position plays much more important role in appropriate processing by computer. In 2001, we will focus more on additional information rather than adjusting the glyph shape from a aesthetic point of view.

In terms of conversation of linguistic and cultural heritage, we believe AA Font should be available in original bitmap as well. As part of this year's achievment, we created  tables of glyphs that are transferred from AA Institute. The tables are accessible at the following web page.


[ Contents | 3. | 5. ]