bugFree UCS Outline Fonts - Bugs: bug #38707, Serbian glyph for CYRILLIC BE + JE...

 
 

bug #38707: Serbian glyph for CYRILLIC BE + JE kerning glitch

Submitted by:  Rosella Capriotti <rosy58>
Submitted on:  Thu 11 Apr 2013 03:45:35 PM UTC  
 
Category: individual character(s)Severity: 3 - Normal
Item Group: letter positioning poorStatus: Fix posted
Privacy: PublicAssigned to: Steve White <Stevan_White>
Open/Closed: OpenRelease: 2012-05-03

Add a New Comment (Rich MarkupRich Markup):
   

You are not logged in

Please log in, so followups can be emailed to you.

 

(Jump to the original submission Jump to the original submission)

Tue 16 Jul 2013 03:32:37 PM UTC, comment #40:

OK I shall do so, actually I'd asked you to change the name because I thought this one had become a general purpose Cyrillic kerning thread. But I see you see things differently.

Rosella Capriotti <rosy58>
Tue 16 Jul 2013 02:58:33 PM UTC, comment #39:

Hi,

You were right on every time. I acted on each comment, except the overall letter spacing of italic vs bold italic.

Notice there is a difference between letter spacing and kerning. Letter spacing has to do with the widths of the glyphs, whereas kerning is a later fine adjustment based on pairs of glyphs. I'm afraid in the case of italic vs bold italic, you're referring to a choice of letter spacing.

If you see further specific kerning issues, could you open up a separate report for them? This has strayed pretty far from the original problem reported.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Tue 16 Jul 2013 01:31:20 PM UTC, comment #38:

The space left between ES and TE looks problematic, seeing that kerning is so tight among the other letters.

(file #28596)

Rosella Capriotti <rosy58>
Tue 16 Jul 2013 01:08:58 PM UTC, comment #37:

As far as Italic is concerned, it seems to me kerning is too tight, especially as compared to Bold Italic.

However, these are just my humble impressions, maybe it's just an optical illusion, let me know what you think about it.

(file #28595)

Rosella Capriotti <rosy58>
Tue 16 Jul 2013 01:03:14 PM UTC, comment #36:

The kerning of Bold Italic Small U+ME and DE+ME differ even though the first letters (respectively U and DE) are similar in shape.
Also, the kerning of Bold Italic Small NJE+I and FE+I don't convince me, I think there is too much kerning.

(file #28593, file #28594)

Rosella Capriotti <rosy58>
Mon 15 Jul 2013 09:16:19 AM UTC, comment #35:

Yes, it's fixed now, I built the font on my own following your instructions, but my humble opinion is still that changing the title would be good.

Rosella Capriotti <rosy58>
Sun 14 Jul 2013 07:35:06 PM UTC, comment #34:

Rosella,

What I forgot, was that the 'em' in FreeSerifItalic is decorated like 'el' on its left side. This is now fixed: in SVN.

Regarding re-naming... the name doesn't need to reflect all the steps taken. The steps ought to be detailed in the SVN logs, which also refers to the bug reports. It's usually better to leave the names alone. Sometimes I change them when they're first reported, if they're quite incorrect, or just too long, or something.

Thanks!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Sun 14 Jul 2013 06:15:55 PM UTC, comment #33:

By the way I think you'd better change the name of this topic to "Serbian Cyrillic kerning" or something like that, as you've performed an overall tune-up.

Rosella Capriotti <rosy58>
Sun 14 Jul 2013 05:07:58 PM UTC, comment #32:

Hello Steve,
It seems you forgot to fix the kerning glitch kerning exposed in file #28518. The new file shows the glitch is still there.

(file #28569)

Rosella Capriotti <rosy58>
Sat 13 Jul 2013 05:01:10 PM UTC, comment #31:

Hello Steve,
The qualms are, as you said, that it wasn't kerned. But so wasn't the Latin "J", so I thought it was because of the "box" issue, you remember we talked about that in this thread a couple of months ago? There's nothing more to add.
And about the praise, you fully deserve it! The other two direct competitors for Cyrillic support, Liberation and DejaVu are no match for you, and what's more, even proprietary fonts like MinionPro or Cambria can't boast of all the fine tuning you've worked out. Maybe, just maybe, you have to pay 100 bucks+ if you want to get all this, and there's still no guarantee of getting it, glitches and outright bugs are always lurking even in expensive fonts.
So, Steve, again hats off! Actually, mere praise isn't enough, all the features and quality I get from you really deserve some $. Where can I donate to the FreeSerif project? I'll gladly do so!
Regards,

Rosella Capriotti <rosy58>
Sat 13 Jul 2013 04:42:24 PM UTC, comment #30:

Yes, the ef had too much space on the right. I'll fix that.

Please describe in detail your qualms with the je.

The product is improved by the work of a lot of contributors,
including problem-reporters. Really this feedback is invaluable.

And thanks for the gushing praise. It's nice to hear sometimes!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Sat 13 Jul 2013 04:26:18 PM UTC, comment #29:

In the zoomed part of the attached image, isn't there too much space between Cyrillic "FE" and "E" especially when you compare that with the tight kerning between "E" and "J"?

(file #28562)

Rosella Capriotti <rosy58>
Sat 13 Jul 2013 04:11:53 PM UTC, comment #28:

Another thing: me too, I'd had some qualms about the kerning of Cyrillic "j", but I declined to bother you when I realised it behaved exactly in the same way as Latin "j", so you should now modify the kerning of Latin "j" because the two letters are homoglyphic (is it the right word?).
Anyway, wonderful job, maybe there are just some glitches here and there, but just minor stuff, overall, I repeat it, the Cyrillic FreeSerif has attained such a quality that even very expensive fonts struggle to reach.

Rosella Capriotti <rosy58>
Sat 13 Jul 2013 04:08:18 PM UTC, comment #27:

Wow!!! Hats off to you! FreeSerif is now beyond any doubt the best open-source Cyrillic font available right now! Even very expensive fonts will struggle to provide the same amount of fine tuning! Seriously, you should charge for it! :)

Rosella Capriotti <rosy58>
Sat 13 Jul 2013 02:45:53 PM UTC, comment #26:

Rosella,
I just posted a snapshot with changes regarding this report.

After a couple of evenings of testing, I was unable to find a definitive change in kerning on Windows due to kerning Unicode against non-Unicode letters. (Kerning is very spotty on Windows, and easy to break. Other fonts also have funny problems in specific apps.) So I went ahead and kerned against the Serbian-style glyphs.

Also, it seems kerning hadn't been well tested in Serbian text generally: three special characters weren't being kerned at all (ju, dzhe, tshe). It looks much better now, to my eyes. Have a look!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Tue 09 Jul 2013 03:48:46 PM UTC, comment #25:

Hello Steve,
May I suggest you also take a look at the combination Serbian (Bold) Italic "de" + "a"? Also "de" + "o" doesn't convince me. Do you want me to typeset a specimen with all combinations both to the left and to the right? Maybe that could be of help to you. Let me know.
Regards,

(file #28523)

Rosella Capriotti <rosy58>
Tue 09 Jul 2013 03:16:58 PM UTC, comment #24:

You're right. The letters need to be closer.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Tue 09 Jul 2013 10:47:44 AM UTC, comment #23:

Hello Steve,

What do you think about the kerning of (Bold) Italic Serbian "DE" (see attached image)? I think this one too needs some tweaking. But I might be wrong, of course.

(file #28518)

Rosella Capriotti <rosy58>
Mon 08 Jul 2013 12:04:49 PM UTC, comment #22:

Another note: the warning does not appear when building TrueType fonts.

Is that just a glitch, or does the problem mentioned in the warning when building OpenType fonts really not happen somehow for TrueType fonts? Since it's currently recommended to use the TrueType fonts in Windows, and the OpenType ones elsewhere, maybe there no practical problem with the kerning, only a warning.

Still need to test.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Mon 08 Jul 2013 11:54:59 AM UTC, comment #21:

For the record, the reason I've been reluctant to implement kerning with the auxiliary letters is because, on building the OpenType font with such kerning, FontForge warns like:

"On Windows many apps will have problems with this font's kerning, because 9 of its glyph kern pairs cannot be mapped to unicode-BMP kern pairs"

(The 9 is because I had specified the Serbian be with several sets of characters from the Cyrillic range.)

I do not know what Windows apps are affected, however.
I'll test it with a few of the main ones.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Fri 19 Apr 2013 07:15:49 PM UTC, comment #20:

Hi Steve,

OK, I think the question is settled, I don't think you need my help anymore: you now have all the pieces of information you need.

Regards

Rosella Capriotti <rosy58>
Fri 19 Apr 2013 02:01:19 PM UTC, comment #19:

Hi,

Yes the O de o example isn't kerned, because of the same technical issue with the Serbian de we have been discussing.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Fri 19 Apr 2013 01:55:09 PM UTC, comment #18:

Hi Steve

OK, I fully understand what you're talking about, don't over-kern and don't redraw the CAPITAL CYRILLIC U, the shape is all right, it's just the natural tendency you described to over-kern.

However, in the last pdf, it seems to me also the last one (the one with "Odo") is not correctly kerned: after all if you push the "d" a bit closer to the two "o"s, the "d" still won't get beneath the two "o"s, so this problem shouldn't be an issue here.

Regards

Rosella Capriotti <rosy58>
Fri 19 Apr 2013 07:44:49 AM UTC, comment #17:

Rosella,

Yes, the Cyrillic U is generally a typographic problem, but that is in the nature of the letter.

In your example U (Serbian)de u, this is the same technical problem we discussed before.

But the other letters are already kerned as much as I would dare (for normal text). Some of them are in fact slightly beneath the serif of the U. If they were closer, the small letter would appear to be actually beneath the U, and this must not happen.

Is this what you're referring to?

I suppose you know, kerning is a very tricky business. It is quite subjective in nature. After a great deal of experience, it amounts to no science, but a set of psycho-visual rules of thumb. But at first, there is a natural tendency to over-kern, resulting in cramped and clumpy-looking text.

The example with the U illustrates this: there is a lot of area between U and de -- how much to kern? You might think to kern them until they almost touch, until the arm of the Y is in fact over the de. But from another perspective, this is far too close. The tops of the letters interfere, while there is still a great area between the letters.

If it's really a severe problem, the proper response is to design the letters, so that such issues are not so severe. I could imagine a U that is weighted a bit more to the right. But I'm not to the point of a re-design of Cyrillic yet...

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Thu 18 Apr 2013 10:57:21 PM UTC, comment #16:

To sum it up, it seems that Capital Cyrillic U too has some trouble with positioning, whichever letter one puts after it, and in whichever style (italic, normal bold etc...)

Regards

Rosella Capriotti <rosy58>
Thu 18 Apr 2013 10:37:21 PM UTC, comment #15:

Hi Steve,

I did some more tests and I discovered boldface too has some issues, and it seems troubles are not limited to Serbian glyphs, take a look at the pdf, first and last one are certainly bad, but the middle ones (sample standard glyphs) look bad to me, what do you think?

Regards

(file #27908)

Rosella Capriotti <rosy58>
Thu 18 Apr 2013 04:04:53 PM UTC, comment #14:

Hi Rosella,

Yes, this is the same issue.

(I wouldn't kern them as tightly as in your example, but a bit more kerning would certainly help.)

I still haven't had time to determine why FontForge makes the complaint...

Thanks!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Thu 18 Apr 2013 01:23:31 PM UTC, comment #13:

Hi Steve,

I was doing some routine check of Serbian kerning and I've just discovered another glitch with CYRILLIC CAPITAL U+SERBIAN BE+CYRILLIC U.

In the pdf you'll see the original (unkerned) output then the modified output with the kerning I suggest: "У\kern-.2exб\kern-.1exу"

Regards

P.S.
Maybe you'd better change now the title to something like "Serbian glyphs kerning issues"

(file #27906)

Rosella Capriotti <rosy58>
Tue 16 Apr 2013 11:56:16 AM UTC, comment #12:

Hi Steve,

They aren't of much use because a web browser is for casual reading, not quality typesetting, yet, ironically, FireFox has a complete support for OpenType features, from localization up to weird ligatures, swashes and the likes.

And they do work, I guarantee I can see Serbian glyphs on Wikipedia and Google Translations.

To be honest, LibreOffice has recently added support for standard ligatures, it's not much but at least better than seeing the edges of the F collide with the point of the I (sorry I don't know the technical terms, I hope you still got the point).

Regards

Rosella Capriotti <rosy58>
Mon 15 Apr 2013 11:03:29 PM UTC, comment #11:

Hi Rosella,

Why are they of not much use in FireFox?

I have not checked to see if they work in FireFox since their font software was changed from Pango to Harfbuzz. (I guess I should.)

LibreOffice -- I know. Believe me, I know. It's a long and dull historical issue. Basically word processors (including the big one) have had terrible support for fine typesetting.
You know what's worse: the mid-range page-layout programs are only slightly better... they are ignorant of most OpenType font features. They've lost touch.

XeLaTeX has for the past few years been the only free software, and the only moderately-priced software, that supports these features, as far as I know. Maybe this will change if Harfbuzz is implemented more widely.

Macedonian in FreeSerif: yes. The table that does this is activated for cyrl{MKD ,SRB }.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Mon 15 Apr 2013 10:29:48 AM UTC, comment #10:

Hi Steve,

As far as these special forms are concerned, in the Linux environment, ironically, Firefox, where they aren't much of a use, does support them, while LibreOffice ... does not! That's why you won't hear much about them unless and until LibreOffice starts supporting them.

Yes I do have another suggestion, did you know the same glyph sets apart from optional italic SHA applies to Macedonian as well? I can't check for myself because XeLaTeX doesn't support Macedonian :(

Regards

Rosella Capriotti <rosy58>
Mon 15 Apr 2013 12:49:14 AM UTC, comment #9:

Hey, thanks!

However, it's kind of complicated, and I doubt that the problem arises on the environments you listed -- I would guess it's a Windows problem....

I can only imagine testing it myself. Don't know when I can do that.

I'm really pleased that somebody is actually using these special forms -- they were added at some pain, as it happened.

Don't hesitate to let me know of any other problems or suggestions!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Sun 14 Apr 2013 11:42:43 AM UTC, comment #8:

Hi Steve,

OK, now I understand you have to maintain the maximum range of compatibility as possible.

Anyhow, I do use Linux and Firefox and Pango, so, if you want me to do some tests for you, be my guest!

Greetings

Rosella Capriotti <rosy58>
Sun 14 Apr 2013 11:19:37 AM UTC, comment #7:

Hi Rosella,

I am somehow not communicating the nature of the problem.

I'm sure it does work in these fonts for you. I have never seen it fail to work myself.

The problem is, when building the font, FontForge warns me that it will fail in some systems. It is not more specific.
(You might ask, why don't I just ask the developers -- that's another story.)

My main targets are Linux (FireFox expecially, the renderers Pango and Harfbuzz, and XeLaTeX. I also do for LibreOffice what I can...) and Windows 7, Vista, and 95 -- but also now WebFonts, which again are another kettle of fish.

When I get a chance, I'll test kerning, despite the FontForge warning) in these systems, but -- it's rather time-consuming.

Cheers!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Fri 12 Apr 2013 07:58:29 PM UTC, comment #6:

Hi,

Yes, that's what I meant, kerning a non-unicode glyph with a unicode one. Minion Pro and I think, LiberationSerif do have kerning for alt/locl glyphs but I've never witnessed any specific issue with that.

Do you want me to try some combination so that I can show you the results?

Greetings

Rosella Capriotti <rosy58>
Fri 12 Apr 2013 07:26:49 PM UTC, comment #5:

Hi,

Let me clarify: FreeSerif is extensively kerned, also in the Cyrillic range.

The issue here is that this specific glyph, the Serbian 'be', is not Unicode, and some problem is reported (whose nature I don't fully understand) with kerning a non-Unicode glyph against Unicode ones.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Fri 12 Apr 2013 02:49:29 PM UTC, comment #4:

Hi Steve!

You're welcome!

What I can tell you is I know for certain there are other fonts (like Adobe MinionPro) which have that special kerning stuff but still work fine in environments which aren't aware of OpenType features. Like Windows, or LibreOffice.

Hope it helps,

Rosella Capriotti <rosy58>
Fri 12 Apr 2013 02:10:18 PM UTC, comment #3:

Hi Rosella,

The question is not whether it will work in LaTeX or most other environments--I too know it will work there. The question is, will it break something else important (such as, overall functionality in Windows).

I'll just have to do some tests.

Thanks very much for your feedback -- only one or two others have remarked on the Serbian support.

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Thu 11 Apr 2013 11:49:26 PM UTC, comment #2:

Hi Steve,

Normally LaTeX works fine with kerning of alternate glyphs, you should include the exception in the OpenType stuff, I don't know how it works, but I'm sure it works :)

I'm sorry I can't be of much help here, I'm a rookie as for as fonts are concerned.

Thanks you very much for your precious interest,

Rosella Capriotti <rosy58>
Thu 11 Apr 2013 09:02:22 PM UTC, comment #1:

Greetings Rosella,

I see it and agree it's a glitch, but I don't know what to do about it.

The technical problem is this:

Kerning is specified by means of a tables that relates pairs of glyphs in the fonts with a relative horizontal offset.

The Russian Cyrillic 'be' lies in the Cyrillic Unicode range, but so far as I know, there is no Unicode character corresponding to the Serbian 'be', as it is regarded as a stylistic variant. This is all well.

However, the font editing software we use, FontForge, warns that kerning will not work in some environments if a kerning pair is not composed of Unicode characters.
This is why these letters are not kerned.

However, I don't know if the warning is correct, or what environments they are, that have trouble with kerning non-Unicode glyphs.

I'll think about it. But I would like to understand the dangers before I implement a kern pair here.

Let me know if you have further info!

Steve White <Stevan_White>
Project AdministratorIn charge of this item.
Thu 11 Apr 2013 03:45:35 PM UTC, original submission:

Hi there,

It seems to me that the SERBIAN CYRILLIC BE + JE combination features a kerning glitch.

I sent you a pdf sample where you can see the difference between the same word with a Serbian BE (the first one) and with the Russian BE.

I'm sorry I don't know how to find out the version I'm using, all I know is that it's the standard version included in Ubuntu 12.10.

I wish to thank you in advance for your attention.

Greetings,

Rosella Capriotti <rosy58>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach File(s):
   
   
Comment:
   

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -unavailable- added by Stevan_White (Posted a comment)
  • -unavailable- added by rosy58 (Submitted the item)
  •  

    Do you think this task is very important?
    If so, you can click here to add your encouragement to it.
    This task has 0 encouragements so far.

    Only logged-in users can vote.

     

    Please enter the title of George Orwell's famous dystopian book (it's a date):

     

     

    Follow 16 latest changes.

    Date Changed By Updated Field Previous Value => Replaced By
    Tue 16 Jul 2013 01:31:20 PM UTCrosy58Attached File-=>Added Italic Small ES+Serbian TE.png, #28596
    Tue 16 Jul 2013 01:08:58 PM UTCrosy58Attached File-=>Added Too much kerning?.png, #28595
    Tue 16 Jul 2013 01:03:14 PM UTCrosy58Attached File-=>Added Bold Italic Small Cyrillic U+ME.png, #28593
      Attached File-=>Added Bold Italic Small Cyrillic Serbian DE+ME.png, #28594
    Sun 14 Jul 2013 05:07:58 PM UTCrosy58Attached File-=>Added Capture du 2013-07-14 19:05:52.png, #28569
    Sat 13 Jul 2013 04:42:24 PM UTCStevan_WhiteStatusProceeding=>Fix posted
    Sat 13 Jul 2013 04:26:18 PM UTCrosy58Attached File-=>Added Capture du 2013-07-13 18:23:35.png, #28562
    Sat 13 Jul 2013 02:45:53 PM UTCStevan_WhiteStatusNeed info=>Proceeding
    Tue 09 Jul 2013 03:48:46 PM UTCrosy58Attached File-=>Added Capture du 2013-07-09 17:46:39.png, #28523
    Tue 09 Jul 2013 10:47:44 AM UTCrosy58Attached File-=>Added Capture du 2013-07-09 12:43:29.png, #28518
    Thu 18 Apr 2013 10:37:21 PM UTCrosy58Attached File-=>Added Bold Serbian BE Kerning.pdf, #27908
    Thu 18 Apr 2013 01:23:31 PM UTCrosy58Attached File-=>Added CAPITAL CYRILLIC U+SERBIAN B+CYRILLIC U.pdf, #27906
    Thu 11 Apr 2013 09:02:22 PM UTCStevan_WhiteStatusNone=>Need info
      Assigned toNone=>Stevan_White
      SummarySerbian glyph for CYRILLIC BE + CYRILLIC JE kerning glitch=>Serbian glyph for CYRILLIC BE + JE kerning glitch
    Thu 11 Apr 2013 03:45:35 PM UTCrosy58Attached File-=>Added SERBIAN BE + JE kerning glitch.pdf, #27838

    Back to the top


    Powered by Savane 3.1-cleanup