[Project-ideas] GSoC idea for improving the aesthetic quality of Indian Scripts

[Sankarshan Mukhopadhyay]

On Sat, Apr 20, 2013 at 4:30 PM, Mayank Jha <mayank25080562 at gmail.com> wrote:
> Well I phrased it wrongly and am sorry for that. By referring to the
> problems pertaining to halant, anuswar, chandrabindu and stuff, I meant to
> imply that the standards pertaining to these would be used in the tool which
> we want to create! I havent faced any such issue in the existing fonts like
> MANGAL.ttf, kruti dev etc. because they are well tried and tested.

Let's get a few things worked out. At the very basic level, in order
to aim for completeness, a font would need to adhere to the
code-points off the Unicode chart for a particular script. However,
all the valid glyphs for a script are not contained within a font.
Among other things, that would make the font "heavy". So, there exist
ordering rules in the rendering engine which work with the various
half forms to provide the combination of conjuncts.

Now, the "completeness" of a font is not much of a concern in this
idea. If you are familiar with the UTRRS system
<https://fedorahosted.org/utrrs/>, you'll notice that there exists
tooling to check for glyph coverage.

What we want to achieve is provide a granular report that forms the
basis of (currently perception bias driven) statements like (for
example) "Mangal is a better font than Samyak".

> I was referring to the new fonts which the user wishes to contribute, and
> which needs testing!

Users usually do not contribute fonts. Let's get that straightened as
well. Font foundries and font developers do. In each case, these folks
take care to turn out a reasonably well ordered font. The idea is to
be create a system which can "test" the fonts. There are a couple of
foundries which turn out very pleasing fonts for Indic scripts, it
should not be difficult for you to check their catalog and, glyph
sets.

> Taking the example of http://www.wazu.jp/gallery/views/View_JanaMarathi.html

Indeed. <http://www.wazu.jp/gallery/Fonts_Devanagari.html> would show
you that there exist a wide diversity in "stylized" font types. And,
it is good to have styles and decorative glyphs.

> 1. The kha letter does not have a proper curvature, at the lower lefthand
> corner.
> 2. The ae ki matra on the letter kha is too straight, ie straight as halant,
> this would not only confuse a newcomer into the devnagri script.
> 3. Also there is lack of uniformity in the manner in which matras are
> applied to the various consonants. for rg the choti e ki matra has no
> joining with the upper line, while the o ki matra has a joining.
> 4. The conjugation of half ka and full ka also doesnt seem to be
> appropriate. the joining line is too straight, it needs curve a bit to make
> it look like there is real conjugation.
> 5. The letter ka also does not look nice, as the two sections on either side
> of the middle line seems to be unrelated

When you observed all of the above, your conclusions were drawn based
on a reference font - which font did you use as a baseline?

> Based on the standards set by expert linguists, we can make a tool similar
> to fontQA.

I hope that when you write the proposal you will include the
characteristics against which the tests would be created. The date by
which the window of student proposal submissions into Melange will
open is near. And, I'm sure you've been working on first drafts.

> As far as how should we go about making the tool.
> I have a doubt, when you said we could also extend fontQA with rule sets
> does that mean creating fontQA testsuites?

If you had looked at the FontQA tool it would have been reasonably
obvious that it has not seen enhancements or, contributions in years.
Hence, whether it is actually possible to expand the tool itself or,
it is required to create a new tool that is similar but flexible -
that is part of the work the interested candidate has to do.

/sankarshan

--
sankarshan mukhopadhyay
<https://twitter.com/#!/sankarshan>