Re: A Pictographic system that makes fonts obsolete

I do not think this system could really work. May may work for some
special cases, but is not so universal.

There are two categories of reasons why I do not consider this system as
OK. I'll tell you first of linguistic reasons, then about implementation
(computer) reasons.

I. Liguistics

The description claims that 66,045,188,505,600 different gliffs can be
made. I do trust the number, but i would like to ask who many "usable
glifs" can can me made? By usable, i mean visualy different.

an example is given with CHmQV:GI:PvWxSM* would it look visualy
different if it was CHmQV:GI:PWxSM* ( i removed the "v" in the last
part). while those two caraters are different, i think they would not be
easy to tell from eachother. A human eye would probably consider them as
identical, or as different writing styles of the same letter. Consider
the letters you write with you pen on a paper, with a choak on a board,
or the one someone else would write under the same conditions. They are
probably more different than that.

Therefore i think this huge number of 66,045,188,505,600 is higly
overestimated.

On the other hand, there are many caracters the system can not represent.

Take chinese caracters, for instance. it is not uncommon to have 5 or
more horizontal lines stack up in a single caracter. The same goes for
verticals. and many more details could not be fiured by a grid of 5*5.
Even a grid of 10*10 might be a bit small. I don't think one could be
confortable designing chinese caracters with a gird smaller than 15*15
or maybe bigger.

the system also alows to change the sytle of line endings to change the
style of caligraphy. But again, if you look at chinese caracters, line
endings can be straight or hooked in the same sytle. In a certain
caracter, some lines must me hooked, other must me straight. To fix
this, have two possibilities. either you increase the grid so much than
you can draw those line endings, or you add new symbols like * to
indicate line endings' style. but you'd have take into acound every kind
of possible ending, so nothing is missing.

What i am trying to point out is the following:
It is true to say that any language needs a finite amount of caracters.
even if this amound can be very high, in case of ideographic language (a
couple of tens of thousands in chinese).

But the number of possible drawings in a limited portion of the plane
(ie a caracter's graphical apearance) is infinite.

Moreover, one caracter is not a single drawing, but a range of
possibility, identified by humans as a single drawing. Take the letter
"i" for instance if the point is higher or lower, you will still
identify the letter. A letter with or without serifs is identified as
the same letter by the reader.

Considering this, the approache used by fonts+codepage is very
reasonable: We have a countable set of abstractions called letter, let's
identify each of them by a number( this is the codepage part). then, if
we have to represent them, let's provide one of the possible
representation of each letter as an image (this is the font).

The system you propose is actualy not a system to reference and display
caracters, but an image format, that assumes it is sufficient to
represent every letter (which I doubt).

II. The implementation problems

But let's assume for a while that the system can accurately describe the
graphical appearance of all the letters you need. Is it efficient?

take the letters you give as an example. CHmQV:GI:PvWxSM*, and
CW:VX:HiNsRqLgH. they take 16 and 15 letters to reprensent. each letter
is stored in one bite (could be less, i know, but it would only
moderately change the results). Let's asumme it is a average number. If
you compare to codepage+font system, it is hudge. Most language's letter
fit in one byte. with 2 bytes (as used in unicode), you can encode more
than 65 000 caraters. Documents based on you system would be 16 times
bigger thant documents in english. 8 times bigger than in chinese.

You could argue that the font system does not only take one byte, since
you have to get the font file too. Ok, but however big the Font file,
from a certain number of caracter, files described by your system will
get bigger. The core reason of this is you did not factorize
information. No matter howoften a letter appears in your text, you have
to describe it completely each time.

Even if typographic variations are possible, they are more limited than
what is possible with fonts. They are closer to chosing between italic
and bold, than chosing between arial and Times new roman. Many people
would find it strongly restrictive.

The system, desipte these consideration on size, could be useful if it
was universal. In that case, you would never need to download any font,
and your system could handle any text with it. But I do not think it is
universal. You'll get a hard time using it for Chinese, Korean, and others.

So it is desing for use mainly (only?) with piktok. So for the end user,
it does not improve any thing compared to a font sytem. The day he
discovers he need to display piktok, instead of downloading the font, he
downloads the software. Moreover, since this is far from standart, I
don't realy see how you can make it work with all existing software. If
people are to get special version of every software for piktok, it don't
think it will be much of an improvement.


All that said, I do not advise you to forget about you system. I believe
It can be of some use. But in an other perspective. Not at all in
replacing fonts. It could be usefull to make up Glyphs.

Here is my point. Any writing system is shaped by the technical means
used to represent it. Writing systems design to be carved in rock, have
different style than those writen with ink on paper.

Restricting you to the system you descibed could give a certain
coherence and realism to your writing system. I don't mean create glyphs
on paper and make them fit in the grid later. I mean design them on the
grid, with the grid.


There was a independant idea on you web page. Regarding the imput of
glyph. You said to type it in roman alphabet, and the system convert it
to the glyph, asking you via a pop up to choose the right one if there
are several glyph matching. This is exactly the way chinese, japanese,
and korean input work. (Maybe you already know it and took the idea from
there). this kind of interface  called IME is standard (at least in the
windows world).If you are to spend time on software solutions for your
langauge, I would advise you to come up with a font, and an IME.

I hope my views on all this are of some help to you.

Florian

Gary Shannon wrote:
> Toying around with a different way to tell software
> how to draw a pictographic character:
>
> http://fiziwig.com/ptgylph2.html
>
> Fonts may not be the right way to go for pictographic
> languages since they were initially designed for
> alphabetic languages, and have to be coerced into
> fitting ideograms or pictograms.
>

Reply