Cyrillic Kern Pairs?

anonymous's picture

Hi All-

I am venturing into new territory and was hoping someone might have some pointers. I am working on a Russian typeface and along with my dim understanding of encodings, I am also at a loss to find relevant information regarding kern pairs. Any relevant caches of information out there that folks can point me to? Ideally, I'd like to find a list of common kern pairs.

Thanks a bunch!

Paul Platosh

hrant's picture

Thank you John! Useful to me too.

BTW, here's a couple of online frequency calculators:
1) http://www.ultrasw.com/pawlowski/brendan/freqcalc.html - But it's no longer working! :-(
2) http://www.volcanokit.com/volcanokit2/ligCounter/ - But it's Latin-only. :-/

hhp

hrant's picture

BTW, I contacted the guy in #1 above, and it turns out he's simply moved it:
http://cgi-bin.spaceports.com/~pharazo/freqcalc.html

Looks really useful, since it also calculates beginning/ending letters, up to tetragraphs, and has nice toggles for various languages and such. But no non-Latin support. :-( The code is there, though, so would anybody be interested in making it Unicode? That would rule.

hhp

eomine's picture

What's needed to make it Unicode-compatible?

BTW, your new link doesn't work for me. The old one is still working though.

hrant's picture

> new link doesn't work

Yeah, same here! Copy-pasting works though. Strange.

hhp

komitlak's picture

macedonian only


text/plainmacedonian kern pairs
mk.txt (1.7 k)


text in unicode
I'll do them all (cyrillic) this summer.

g

hrant's picture

BTW, the script for that page above is now available:
http://ultrasw.com/pawlowski/brendan/frequency.txt

If somebody is willing to make this Unicode, that would be super sweet.

hhp

eomine's picture

It was already available when you posted it one week ago.

The characters' counting algorithm is there. What kind of technology/feature is required to make that script Unicode-compatible? IIRC, Python can deal with Unicode values.

komitlak's picture

so far:


application/octet-streamobid_06
03__Letter Frequency Calculator.rar (3.0 k)


needs python, wxPython 2.4 and PythonCard

hrant's picture

So if I have FL4.6 + Python (which version?), what should I do?

hhp

eomine's picture

Looks like it is a Mac file.
Could't you please just post the code (pure text), Komitlak?

I'm guessing we don't need FL to run this script.

komitlak's picture

I apologie for not providing any explanation. The file is rar compressed, and it can be uncompressed with Winrar, Winzip, WinAce, Total Commander (Alt+F9, this is ex Windows Commander thnxs to M$).
Here are some links:
http://www.rarlab.com
http://www.winzip.com
http://www.winace.com
http://www.ghisler.com (generally this is a musthave)

The app needs no FL, and I believe it will work fine with any Python 2.x version, but be shure to have the adequate version of wxPython and PythonCard (must have both). Theese are free for use and download:
http://www.wxpython.org/download.php#binaries
(Windows, Mac OS X, Linux...)
http://pythoncard.sourceforge.net/installation.html
(Windows, Mac OS X, Linux)

Finally there are two files in the rar archive: obid_06.py and obid_06.rsc.py. Double click on obid_06.py and after a few seconds (or minutes :^) the app will launch a window that imitates Brendans calculator, with lot of not-working-at-the-moment text-fields, radiobuttons and checkboxes, except the really important one - the big input text-field. Actually i blindly copied his calculator, but didn't figured all the meanings of his parameters (ie frequency), and added a codepage selection list wich is probably unnecessary.

Now grab some text and paste it in, click the

hrant's picture

I have WinZip8, and it won't load it up. :-(

And just to check: you make the script Unicode-savvy, then?

hhp

komitlak's picture

Maybe this should have it's own tread.

Hrant i'm sorry, i'll never (close future) post here a rar file again. About the unicode, the script doesen't care what is it working with, and so far i haven't done anything to mess that up :^) it works fine with cyrillic, see the picture below.
I'm thinking to make an outside unicode txt file that will contain all the characters that we dont want to be treated as letters (numbers, punctuation...) so every user can define what he/she wants to be counted and what not.

Here are two screenshots

screenshot 1
screenshot 2

Here's the source.


text/plainsource
obid_08.py (6.8 k)



text/plainwindow sheme
obid_08.rsrc.py (6.1 k)


If you already have installed Python, wxPython and PythonCard, download obid_08.py and obid_08.rsrc.py in one same folder, and then just doubleclick obid_08.py

As i'm aware there is no windows specific coding here, so the script should work on Mac and Linux too.

Hope you'll have no problems this time, and please send your opinions.

hrant's picture

> About the unicode, the script doesen't care

Really? Because I tried to paste in some Cyrillic text (part of the second file from John's zip) into the original website version, and it spit out junk... But I'll try yours tomorrow - it sure looks great in those screen grabs! Thanks.

hhp

hrant's picture

Both the wxPython and PythonCard installers gave errors (similar or identical ones), after first opening up a command window.

I have Python 2.2 installed - because I think anything higher throws off FontLab, no? And I don't want to cripple my FontLab (version 4.6.0).

hhp

eomine's picture

It seems to be working fine here!

hrant's picture

Which version of Python do you have installed?
If it's greater than 2.2, does it still work OK with FL?

Komitlak, could you somehow make a stand-alone utility of this thing?

hhp

eomine's picture

It's 2.2 here. About FL/Python compatibility issues, I know nothing.

Did you get the correct versions of wxPython and PythonCard? You'll need the wxPython for Python 2.2, with Unicode support. After installing wxPython, run the installation script for PythonCard.

hrant's picture

Yup, that's exactly what I did...

hhp

eomine's picture

What error are you getting there?
BTW, do you know Perl? Because it seems that Perl has Unicode support too.

John Hudson's picture

Hrant, the excellent free Power Archiver (www.powerarchiver.com) will uncompress .rar files and most other formats.

Sorry I have not posted my FontLab Cyrillic kerning files yet: it's a bit complicated because they use my production names, so you need a whole other set of files to make them useful.

hrant's picture

A black command window opens up, then I get:
"
C:\DOCUM~1\HRANTH~1\Desktop\SKG2\LETTER~1\WXPYTH~1.EXE
The NTVDM CPU has encountered an illegal instruction.
CS:053c IP:0119 OP:65 63 74 20 61 Choose 'Close' to terminate the application.
"

Maybe I need to reinstall Python? If so, should I go with 2.3 instead? I worry about FL...

I used to do PERL, but I've become lazy with age...

hhp

komitlak's picture

Hrant, i have no idea what can be the problem. I'm working with Python 2.3 but Eduardo says that it works fine for him with Python 2.2!

>> Komitlak, could you somehow make a stand-alone utility of this thing?

i made one but i don't know if it will work on another PC, i'll post it in the next 24 hours. Can't do that now cos from where i'm it will upload thill the end of time (5Mb).

i think the best way to do this is with C/C++/C# or similar, but that's out of my leage.

g

komitlak's picture

After an unplanned trip to the country:
i've tried to upload the stand alone app but i get an error (maximum allowable upload size).

don't know what to do

hrant's picture

Do you have a server you can upload it to and then provide a link?
If not, just email it to me* and I'll do it for you.

* hrant-thatsymbol-inverselogic-dot-corn

hhp

komitlak's picture

we have limited space on elfasko org but i'll try

g

ps Hrant, please finish the trapping text on your site, it's wery educative.

anonymous's picture

John-

You are a god send! Thank you for the info, it is much appreciated. I'd love to get the Fontlab metrics lists, since I think that would help as well.

THanks once again,

Paul

John Hudson's picture

I don't have a list of common pairs, but I do have an extensive list of Russian words that I use to test spacing and kerning of Cyrillic fonts. These are derived from online Russian dictionaries and word lists. The idea is to show at least one example of every combination of two letters that occur in the language. Here they are:


application/x-zip-compressedRussian kerning
Cyrillic Kerning source files.zip (17.7 k)



Note that all these words are Russian, so I can't guarantee that a font kerned using these will work ideally for other Cyrillic-script languages. There may be, for example, combinations of letters that occur in Ukrainian that do not occur in Russian.

If you are interested, I also have versions of these lists designed for use in the FontLab metrics/kerning window, i.e. using the /GLYPHNAME convention instead of the Cyrillic characters. However, the glyph names I use are Tiro's internal production names, not the AFII or uniXXXX names recommended by Adobe for the shipping font. We have a rather elaborate glyph name management system that allows us to switch back and forth between production names and final font names. If you're interested, I can fit you out with these files too.

Syndicate content Syndicate content