Merde! That was a brain-bender.
Basically, you can't get to where you want to go from where
you're at now.
The problem is somewhat comparable to calligraphic Arabic (also
an alphabetic writing system. Look at the calligraphic alJazeera
logo and try to imagine the font that could create it. The chars
are a-l-j-a-z-i-r-a. There is no such font. Despite the beauty
of Arabic calligraphy, fonts for mundane use are nearly all
written to a baseline and look rather ugly and mechanical, which
they are, to the native reader. But that's as much as can be
done with current technology.
Unfortunately for you, Unicode came about in part because there
really is no display driver in use today (that I know of, anyhow)
that "paints" bits and pieces onto the screen to compose each
glyph. So all chars in all fonts, pretty much, no matter whether
they're simple Latin chars with diacritics or godawful Hanzi
chars combining 8 different radicals, are complete in every cell.
Which is why a basic real-Hangul font is about 11000 chars! (Out
of curiosity, why didn't you just grab glyphs from a real Hangul
font? Or was it for that very reason: you took one look at that
11000 lower bound and decided "not in THIS lifetime!"?)
Unless you want to write your own display driver, the best you
can do is follow the Hangul model which looks like a 2x2 square
for most glyphs. Since English is alphabetic rather than
syllabic you can get away with a 1Wx2H model. Standard display
drivers expect to advance their pointer across the screen after
every char, so that constrains your choices pretty completely, as
it does everyone's.
One choice is to carry on as you're doing, but shrink your cell
size til there's nothing left on any side, and then create
several space chars to use wherever needed. Then you could
"paint" by consuming more than one editor line for each
composite-glyph line, moving up and down between words. The
words would have to be output as images, not as characters.
Another choice, closer to real Hangul, would be to create all
possible vertical combinations (the 1Wx2H cel), and then
carefully figure out each time how to combine them to represent
the words you want. So to write "trains" in the way that sample
was composed, which was (if spaces get thrown away, this likely
won't look so good to you as it does now to me)
t r top-of-i
a1 a2 middle-of-i
n s bottom-of-i
But that's a 1x3 cell, not 1x2. Using 1x3 cells would
definitely create an overly-complex font for not much benefit,
though, and 2-char and odd-count words would always cause
problems in composition. Your "trains" word, being even-count,
should have been, for ease in designing and writing, based on a
1x2 cell making "trains" a 3x2 word,
t a n
r i s
or
t r a
i n s
The cell count would be (26 x 26) + 26 with all odd chars being
represented vertically by their own 1x2 cell. Trying to include
more freedom would increase the cell budget exponentially and,
again, for not much benefit. I certainly wouldn't do it because
I'm sure there'll be enough visual interest without more freedom
than you already get with 26x26+26.
The nice thing is that using such a font doesn't require
displaying the output as an image.
It's not sorted into a 9*9 square, but all of them fit a set of
guides. I've attached my SFD and an SVG containing all the guides for
the rest of the characters. At the moment, only the glyphs for the top
of each block have been put in the font, once they're working it
should be easier to make the rest work.
By your description, it should work if you have it sorted into a
9x9 square (I think Korean also uses 9x9, as Chinese and, for
Kanji, Japanese do).
So if it's not working, then you're doing something you don't
realise you're doing (something I experience all too frequently
myself).
Would it be at all possible for you to post your sfd file as an
attachment? Perhaps if I loaded it and looked at it, the problem
would become visible to me--something that's very much not the
case now!
Each glyph as I have it is one of those components in the page. Imagine if
you were to delete all of them bar one and then take the whole
1000px*1000px square with the letter in and that is the glyph. Every glyph
in the font as I have it is 1000px wide and 1000px tall with lots of white
space and only a small path in the upper left or lower right or wherever
for that glyph
Post by Samuel SutherlandAnd in my great stupidity, I attached the wrong link. Attached to this
email is the word trains as an SVG
I managed to load that into Inkscape under BSD, since XP didn't
way to know.
My first thought is that, if the individual components in that
image are also the individual glyphs in the font that you want to
assemble into results like that "trains" image, I don't really
see how you ever make it work. The pieces are very different in
size, something that font software doesn't like font glyphs to
be.
My second thought is that I don't see any superimposition.
Abutment, yes, and even partial overlap, but not superimpostion.
So are those component pieces not really glyphs in your font, or
what?
Post by Samuel SutherlandPost by Samuel Sutherlandhttp://imgur.com/a/lp4RI
This is what the word "trains" should look like.
What I have right now is a differerent glyph for each character in
each position, and each glyph is 1000px by 1000px with the actual
glyph in the right part of the box, so that were all six glpyhs that
would make up "trains" superimposed, this is what they'd look like.
I realise it's quite ambitious to do this with a system that was just
designed for more simple kerning and substitution, but I reckon it
would be possible, if a little difficult.
Post by s***@att.netYou might need to explain the issues in more detail. Can you
fake up an image of what you're expecting to get, using your
example glyphs/fragments, so we can see the difference to what
you are getting? Em is not necessarily a good metric for the
amount to move to get perfect superimposition since in most most
fonts individual glyphs vary in width.
Post by Samuel SutherlandI am currently in the process of recreating a hand-written
pseudo-korean script for the latin alphabet for a friend's dungeons
and dragons campaign, and what this means is that multiple characters
can be joined together in single blocks. This is absolutely possible
to do using kern and calt to replace characters with the correct
variant glyph for the part of the block they're in, and then kerning
the new glyphs together.
To this end, I recreated each glyph in the correct position inside a
1x1em square (1000px because opentype) in inkscape, and imported the
SVGs with no problems. Therefore, in order to put the characters in
the right places, I should only need to kern the second glyph back by
exactly 1em so that the two glyphs are superimposed, with their left
bearings and right bearings aligning. Unfortunately, this is not
working and the actual value that kerns the two glyphs into the right
position is around -833. I've attached a link to screenshot of how
they kern
Post by Samuel SutherlandPost by Samuel SutherlandPost by s***@att.netPost by Samuel Sutherland(roughly) correctly with a value of -833, and also a screenshot
showing them being 1em apart with the calt lookup removed. In this
unkerned image, the vertical line at the left of the 4-shaped glyph
for 'b' on the right hand side should exactly overlap with the right
hand side vertical of the H-shaped glyph for 'b' in the third box
Does anyone know why I need to use -833 over my expected value of
-1000, because -833 seems very arbitrary to me - everything in the
actual glyphs is a multiples of 25px because I built them all in
inkscape with a load of guides to make sure all the characters would
be in the right place inside the 1em square box to make my job easier
now. Oh how naive I was.
Many thanks in advance to anyone who knows what I'm not seeing,
Sam Sutherland
Screenshots: http://imgur.com/a/lp4RI
(P.S. I'm aware that the glyphs aren't too pretty - my priority is on
getting it working over getting it pretty - glyphs can always be
edited afterwards)
--------------------------------------------------------
----------------------
Post by Samuel SutherlandPost by Samuel SutherlandPost by s***@att.netPost by Samuel SutherlandCheck out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
fontforge-users mailing list
https://lists.sourceforge.net/lists/listinfo/fontforge-users
http://fontforge.10959.n7.nabble.com/User-f8781.html
------------------------------------------------------------
------------------
Post by Samuel SutherlandPost by Samuel SutherlandPost by s***@att.netCheck out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
fontforge-users mailing list
https://lists.sourceforge.net/lists/listinfo/fontforge-users
http://fontforge.10959.n7.nabble.com/User-f8781.html
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
fontforge-users mailing list
https://lists.sourceforge.net/lists/listinfo/fontforge-users
http://fontforge.10959.n7.nabble.com/User-f8781.html
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
fontforge-users mailing list
https://lists.sourceforge.net/lists/listinfo/fontforge-users
http://fontforge.10959.n7.nabble.com/User-f8781.html
