Design of Common Ligatures

Discuss FontCreator here, please do not post support requests, feature requests, or bug reports!

Which design works best?

Plain Text
2
33%
Verajja Ligatures
2
33%
DejaVu Ligatures
2
33%
 
Total votes: 6

William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

sevry7 wrote:In yours you need various cells for the separate combinations, which you have in E7xx.

Missing, of course, are the rest of the long S ligatures and triples. While there is ffi, already, there is not yet long S fi, nor f long S i, etc.
Actually, the golden ligatures collection runs from U+E707 to U+E7BF.

The character at U+E7E0 is one of the ligatures from my science fiction story scenario.

http://www.users.globalnet.co.uk/~ngo/ligscifi.htm

I am sorry if I have caused confusion.

The following web page introduces and indexes the golden ligatures collection.

http://www.users.globalnet.co.uk/~ngo/golden.htm

The following page adds some of the ligatures you mentioned.

http://www.users.globalnet.co.uk/~ngo/ligatur3.htm

In addition, I produced the following page: perhaps some of the code points might be the ones's that you are seeking.

http://www.users.globalnet.co.uk/~ngo/ligatur4.htm

Your post has lots of other interesting items as well. I am hoping to reply more fully.

Hopefully these code points will be of interest.

William
sevry7
Posts: 32
Joined: Mon Jan 31, 2005 4:50 pm

Post by sevry7 »

William wrote: Your post has lots of other interesting items as well. I am hoping to reply more fully.
It was previously suggested that the unassigned areas at the beginning of FB could also house ligatures.

I suggested using a wide long S instead of separate ligatures, for combinations of long S followed by letters with left ascenders. A table could be used to make the substitution of the symbol for the regular s, should the flow control find, sb, sf, etc.

'high' Latin: AE, OE, small and cap, and the normal long-S.

FB00 small double F
FB01 small FI
FB02 small FL
FB03 small FFI
FB04 small FFL
FB05 small LONG-S T
FB06 small ST

then

FB07 small CT
FB08 wide long-S
FB09 small double long-S (MUFI EBA6)
FB0A small long-S I (MUFI EBA2)
FB0B small long-S L (MUFI EBA3)
FB0C small long-S F
FB0D small double long-S I (MUFI EBA7)
FB0E small double long-S L (MUFI EBA8)
FB0F small double long-S F
FB10 small double long-S (alt version)
FB11 small long-S H (MUFI EBA1)
FB12 small long-S F I

Arabic might have E8xx in some fonts.

So E7xx might be safe. And that's what you've used.

Much like the Greek and extended, at this point, one is left with a number of long-S/F or F/long-S combinations with one or more letters following. It could get tedious, and the particularly if one needs to have cells composited with 'half-marks', as in the Greek.

MUFI has the OE and OO with marks. AE must be somewhere. So those are 'standard'.

I think the ones above capture most of the common ligatures. But again, if one has to spell out all the possible combinations, and then with marks, that could get tedious. But these might be useful:

small long-S F L
small long-S F F
small F long-S L
small F double long-S
small double long-S T
small I T
small A, E, I, O, U with script S (italic font)
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

> They do include a lot that I've never seen in old documents. But I'm sure they are found, maybe from German printers.

I think that MUFI tries to include characters from manuscript documents as well as from printed sources. My golden ligatures collection was intended just for printed sources, in particular English 18th Century, so that the information of ligature usage could be conserved and displayed in transcripts using ordinary font technology.

What remains? In my case, a cell is needed for the stand-off long S ...

I have not seen the term "stand-off long S" before.

> MUFI suggests quite a variation in old texts, many of which variations I have never seen. And those that I have seen and wished to duplicate were not part of MUFI, though many are included in your Chronicle.

As for the rest, if you would like to state what they are, maybe we can work out some Private Use Area codes for them so that we are both using the same codes.
William wrote: I have come to the conclusion that a central issue is that, as at the present time, with OpenType support not in wide use, that entering transcripts using Private Use Area precomposed glyphs is a good idea, yet that a method is needed so that, in the future, transcriptions to regular Unicode can be made, preferably by an automated method.
> Don't the open type extensions that allow auto-substitution look to Unicode cells, anyway? I haven't set up a font that way. So I don't know.

As I understand it, the answer is that usually no, but they can do. When you say Unicode cells, that is perhaps not the best way to think of it. Unicode has code points, font maps have cells, each cell contains a glyph, or maybe no glyph, such as the space and various specialist spacing characters.

In order for an application, such as a desktop publishing program, to use a TrueType font, a cell needs to be mapped to a Unicode code point. What you term is cell is often just called a glyph, though since you use the term cell, then upon thinking about it, I think that perhaps that is a better terminology as the cell contains a glyph.

With an OpenType font, in those parts where glyph substitution is used, the glyph does not need to be mapped to a Unicode code point as the glyph is selected using the glyph substitution rules which a font designer has added into the font. The glyph can be mapped to a Unicode code point if the font designer chooses to do so, usually a Private Use Area code point in practice, so that people using non-OpenType aware applications can access the glyph, perhaps to produce a display using a desktop publishing package: however, some people are against that practice: however, I happen to think that it is a good idea.
William wrote: meaning of Private Use Area ligature glyphs could be added into the font in the Description section of font, so that in the future the font file could, in principle, be accessed by a special purpose application program which could extract that information and use it for such purposes as producing an OpenType version of the font and for decomposing into regular Unicode any text which had been keyed using the font.
> I would think some kind of table would be needed, even duplicating any MUFI or other 'standard'.

Yes.

> People will surely have unexpected needs, and might need to put various symbols in private area which hadn't been anticipated by others.

Yes.

> But I thought the extended open type features were based on table lookup?

Well, glyph substitution for ligatures is based on table lookup. Please note that that is done with glyph numbers not Unicode codes, presumably because all glyphs in a font have a glyph number yet all glyphs in a font need not be mapped to a Unicode code point.

> If that's not done, then one has to hardcode the Unicode right into the text.

If you mean hardcode the Private Use Area codes for the ligature glyphs into the text, then yes.

> Or are you suggesting that the Unicode be hardcoded in that way, but that table could be used by the app to somehow create a virtual text which is perfectly searchable? in addition to using perhaps another table for auto-substitution.

I am not quite sure that I have understood what you mean here, but I think that the answer is probably no. However, you may have suggested a good idea here as an interim solution until OpenType becomes widely used.

> As I said before, what I do is rely on a particular ordering of regexp to run through the text and replace the Unicode with lower Latin letters. But I have to generate a separate document, which is then searchable.

regexp?

What I was trying to suggest is that if someone is making a TrueType font with ligature glyphs in the Private Use Area, because he or she cannot, or does not in some situation, produce OpenType fonts then the TrueType font could contain notes, in some format which is both human readable and machine understandable, in the Description part of the TrueType font, so that at some later stage the font could, in principle, be converted to an OpenType font using an automated method.

William Overington

1 May 2007
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

I have now uploaded Chronicle Text M version 0.261 to the web.

http://www.users.globalnet.co.uk/~ngo/CHRONTXM.TTF

This is a special version of Chronicle Text 0.26 with the following changes.

Twelve ligature glyphs are mapped to the MUFI 2.0 recommendations and the original mappings removed. This was done by copying the original glyphs, adding the mappings then deleting the original glyphs together with their mappings. This resulted in the glyphs which are mapped using MUFI 2.0 being in code point order above the remaining glyphs which are mapped using the golden ligatures mappings. The MUFI 2.0 mapped items are as follows.

U+EBA1 LATIN SMALL LIGATURE LONG S H

U+EBA2 LATIN SMALL LIGATURE LONG S I

U+EBA3 LATIN SMALL LIGATURE LONG S L

U+EBA6 LATIN SMALL LIGATURE LONG S LONG S

U+EBA7 LATIN SMALL LIGATURE LONG S LONG S I

U+EBA8 LATIN SMALL LIGATURE LONG S LONG S L

U+EEC5 LATIN SMALL LIGATURE CT

U+EEC9 LATIN SMALL LIGATURE FJ

U+EECB LATIN SMALL LIGATURE FT

U+EECE LATIN SMALL LIGATURE FFT

U+EED1 LATIN SMALL LIGATURE GG

U+EED6 LATIN SMALL LIGATURE PP

The science fiction ligatures have been removed, as have the PUA copies of glyphs from the regular Unicode range above U+0100 which were originallyadded to facilitate copy and paste operations in some graphic making circumstances. Also the U+F001 and U+F002 mappings of fi and fl have been removed, the font repostscripted and the glyphs for fi and fl moved to be with the other ligature glyphs in the U+FB0. range.

I have changed the design of f and long s themselves so that a gap will show in ligaturable situations if a ligature glyph is not used. This is rather back to front of the way that metal ligatures were designed. In those cases the metal ligature was to avoid a clash with an overhanging part of an f or a long s. In this font the gap is used so that people can more easily tell that a ligature glyph is in use.

Thus the font has twelve ligature glyphs mapped according to MUFI 2.0 and many more mapped using mappings from the golden ligatures collection. There are several more glyphs that I may be able to move, such as tt and tz, yet I am wondering how to change the designs so that the use of a ligature is noticeable. Also, hopefully I can move M macron, m macron, N macron and n macron when I have studied MUFI 2.0 soem more. Also, I can hopefully add some of the MUFI glyphs which are not in the golden ligatures collection and not in the font at present.

Anyway, here is a black letter font with some MUFI 2.0 mappings for ligature glyphs, so hopefully it might be of interest.

William Overington

1 May 2007
sevry7
Posts: 32
Joined: Mon Jan 31, 2005 4:50 pm

Post by sevry7 »

William wrote:
sevry7 wrote: What remains? In my case, a cell is needed for the stand-off long S ...
I have not seen the term "stand-off long S" before.
Wide long-S. The small long-S H would use a stand-off long-S. A small long-S E would not. That's all. In other words, one uses a different long-S, depending, instead of specifying composite-glyphs/cells for each possible ligature.

William wrote: With an OpenType font, in those parts where glyph substitution is used, the glyph does not need to be mapped to a Unicode code point as the glyph is selected using the glyph substitution rules which a font designer has added into the font.
Again, I would assume it's a table. That is, if "CT" is found, then the Unicode CT cell is associated with it in the table and that it what is displayed when the text is 'flowed'. If a small "SS" were found, the table would point to the proper ligature. And in my case if small "SH" were found, it would point to two letters/glyphs, one for small wide long-S, and one for the small H. Of course, as I have seen alternate forms for "SS", I don't know how alternates would be triggered (perhaps by a tag juxataposed with the letter if one is using HMTL?).

Short of all that, when I said 'hardcoded', I meant that all the Unicode callouts are specified right in the HTML formatted text. That means the text can't be searched, for all the embedded special characters. In my example, I also have an English translation that doesn't use ligatures. But for some of the still standard ligatures, I would rather than it did.

Instead, as a workaround or hack, I use a series of regular expressions (regexp, or regex) to filter the HTML formatted text to remove all of the hardcoded ligatures, placing the result in a new page/window so that it might be searched. So in the example I showed, previously, the filtered resulted would show "CT" instead of the "CT" ligature.

But again, I'd just as soon that the filtered result, itself, retain a few ligatures because these simply look better. With open type extensions this would be possible. Without it, even a few hardcoded ligatures would make the document unsearchable, unless the search phrase itself included the Unicode codes.

William wrote: because all glyphs in a font have a glyph number yet all glyphs in a font need not be mapped to a Unicode code point.
The 'character' is the one thing, and its mapping particularly to a Unicode value is another. I didn't mean to be confusing, if I was, previously.

William wrote: What I was trying to suggest is that if someone is making a TrueType font with ligature glyphs in the Private Use Area, because he or she cannot, or does not in some situation, produce OpenType fonts then the TrueType font could contain notes, in some format which is both human readable and machine understandable, in the Description part of the TrueType font, so that at some later stage the font could, in principle, be converted to an OpenType font using an automated method.
In other words, just explain the idiosyncratic mapping, as a sort of essay embedded in the font, so that conversion to some future standard, as yet unknown, might be made easier? Wouldn't that information be self-evident to someone using a program like Font Creator?

William wrote: Chrontxm
You've made changes. What do you think of the extra 'FBxx' glyphs suggested previously. Those are some very common ligatures.

If one could take those off the table, it leaves, as suggested before:

small long-S F L (E793)
small long-S F F (don't know if this is needed)
small F long-S L
small F double long-S
small double long-S T (E75C)
small I T
small A, E, I, O, U with script S (italic font)

You include, in addition

small double long-F T (EECE)

But then you have a requirement for the others, as well.

What about (you can see the 'script' S in the word, Catechumenus, in the previous example - and IT was suggested by the other previous example in the thread):

EEE0 small long-S F L (E793)
EEF1 small F long-S L
EEF2 small F double long-S
EEF3 small double long-S T (E75C)
EEF4 small double long-F T (EECE)
EEF5 small I T
EEF6 small A small script S (italic font)
EEF7 small E small script S (italic font)
EEF8 small I small script S (italic font)
EEF9 small O small script S (italic font)
EEFA small U small script S (italic font)

I mentioned that there are alternatives, as often found in Greek ligatures and abbreviations. Perhaps alternatives would not merit a Unicode composite, but would simply be an alternative font, or just the bold version of the font, etc. So instead of worrying about an alternate long-S, one just has the long-S, and that's it, and no A script small S, etc, either (though I think this latter might benefit from having it specified in Unicode)?

I could give an example, as well, of some of the greek abbreviations and ligatures I selected, from among MANY, which I placed in 3A5x, 3A8x, and 501x:

Image


As I mentioned in another thread, the grayscale works very well, even in small size fonts. However, it does require that Font Smoothing be enabled in Windows. If it isn't, the font looks awful. I wouldn't mind having the option to hint in Font Creator. It's been suggested, before.
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

I have been adding some more ligatures to the Chronicle Text M font this morning and have uploaded the resulting font to the web.

Readers who downloaded version 0.261 and would like to keep a copy of that might like to rename the copy of CHRONTXM.TTF which they already have to become CHRONTXM0261.TTF before downloading the new version of CHRONTXM.TTF.

The latest version of CHRONTXM.TTF is version 0.262 and is available as follows.

http://www.users.globalnet.co.uk/~ngo/CHRONTXM.TTF

This version has a number of additional ligature glyphs which are mapped as specified in the MUFI 2.0 recommendations.

William Overington

7 May 2007
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

I have been adding some more ligatures to the Chronicle Text M font this afternoon and have uploaded the resulting font to the web.

Readers who downloaded version 0.262 and would like to keep a copy of that might like to rename the copy of CHRONTXM.TTF which they already have to become CHRONTXM0262.TTF before downloading the new version of CHRONTXM.TTF.

The latest version of CHRONTXM.TTF is version 0.263 and is available as follows.

http://www.users.globalnet.co.uk/~ngo/CHRONTXM.TTF

This version has a number of additional ligature glyphs which are mapped as specified in the MUFI 2.0 recommendations.

U+EBA5 LATIN SMALL LIGATURE LONG S P

U+EBA9 LATIN SMALL LIGATURE LONG S TI

U+EBAA LATIN SMALL LIGATURE LONG S TR

U+EEDD LATIN CAPITAL LIGATURE PP

U+EFAD LATIN SMALL LIGATURE OC

Also, M macron, m macron, N macron and n macron are moved from their golden ligature collection mapped places to MUFI mapped places.

William Overington

12 May 2007
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

Some readers might like to know of these mentions of some of the code point allocations in the golden ligatures collection.

http://www.joern.de/tipsn98.htm

http://www.joern.de/tipsn128.htm

http://www.guarraci.de/kbdgr/kbdgrfr.html

http://www.macinplay.de/yabbse/index.ph ... ;topicseen

William Overington

19 May 2007
cxhris
Posts: 5
Joined: Sat May 12, 2007 11:54 am
Location: uk
Contact:

Post by cxhris »

don't know if this thread is still being read, but:
some questions about ligatures, and other alternate glyphs, and where to put them... and does it matter..?
first: ligatures.. i'm testing some i made out in a web browser, using &#xxxx;. i'd made fi fl ffl etc at unicode FB00 or so, and carried on beyond... ci ct gi si gg oo gt fj ... i went through the 12 reserved boxes and into the armenian forms...: the browser displayed them ok to begin with, but then started to give the notdef sign... and i got the same quirk when i put them in this 'golden ligatures' E700 area. so then i mapped them to a low number, starting at 1000, coptic or something, and they display fine.
But: Volt asks you to identify the substitutions by glyph index number. So, does it matter where they actually go...? this is my 1st go at ligatures, so maybe this is a daft question... :oops: it's just that i don't own an up to date programme that can do the switching. :(
and then:
what about swashes and old style numerals and so on. are there preferred areas for all? all in the same place?
Bhikkhu Pesala
Top Typographer
Top Typographer
Posts: 9873
Joined: Tue Oct 29, 2002 5:28 am
Location: Seven Kings, London UK
Contact:

Post by Bhikkhu Pesala »

It does matter where you put alternate glyphs, though within the Private Use Area you can map them any way that you wish. I wouldn't recommend using the Coptic character set. It will confuse users who use your fonts with Coptic characters and it prevents you from adding Coptic characters to your fonts, should you ever wish to do so.

If you're using VOLT then they don't need to be mapped at all (take a look at Palatino Linotype, for example), but as you say that you have no applications that support OpenType features then they do need to be mapped or they cannot be used.

FontCreator assigns the following Private Use Area codepoints for use with Small Capitals, Stacking Fractions, Old Style Numerals, and Ligatures:
58033-58126 Basic Latin Small Capitals
58161-58255 Latin-1 Supplement Small Capitals
58256-58383 Latin Extended-A Small Capitals
58384-58591 Latin Extended-B Small Capitals
58912-58974 Basic Greek Small Capitals
59072-59218 Cyrillic Small Capitals
59680-59829 Latin Extended Additional Small Capitals
60531-60542 Number Forms, Nut Fractions or Stacking Fractions
61125-61148 Selected Ligatures from the Medieval Unicode Font Initiative
All values are in decimal notation since the Complete Composites feature data file uses decimal codepoints.

The Medieval Unicode Font Initiative is an extensive proposal to the Unicode Consortium for inclusion of additional codepoints. FontCreator includes definitions for some of these. It may save some work to use the Transform Scripts to add the ligatures to your fonts, then the mappings will be done for you.

See my tutorial on Customising Unicode Data for some insight into how the Private Use Area is currently allocated.

The standard Unicode data text file distributed with FontCreator does not include these non-standard definitions in the Private Use Area. Mappings in the PUA had to be assigned in order to make Complete Composites and Transform scripts work at all, and to use these extended features in applications that do not support OpenType features.
My FontsReviews: MainTypeFont CreatorHelpFC15 + MT12.0 @ Win 10 64-bit build 19045.2486
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

> ... and i got the same quirk when i put them in this 'golden ligatures' E700 area.

The golden ligatures collection is just the independent idea of one person (me!) and the specific code point allocations have no formal validity with Unicode. However, they are Unicode compatible in that they are a particular use of part of the Unicode Private Use Area, the Unicode Private Use Area being available for any person to use for his or her own code point allocations. All allocations within the Unicode Private Use Area are not unique as any person may define such allocations, however they are guaranteed not to conflict with the code point allocations of regular Unicode.

Using codres in the range U+FB07 upwards for ligatures is, in my opinion, a mistake. This is because the codes are not offered for private use allocations and applications may know that they are unassigned and may treat their use as an error.

The use of Unicode Private Use Area code points for ligatures is thought of badly by some typographers, though not by me. Some people seem to take the line that unless one has an OpenType font with an OpenType-aware application that one should do without ligatures. I do not agree. There is a distinction between using such ligature glyph code point allocations for producing display effects locally and using such ligature glyph code point allocations for sending information from one user to another user. The former seems alright always to me. The latter seems to me alright if used with care and the recipient is aware of the code point allocations being used.

I would advise though not to use codes in the range U+F000 to U+F0FF for your allocations as that range was used by Microsoft for its now obsolescent yet still used Private Use Area allocations for fi ligature at U+F001 and fl ligature at U+F002 and for its symbol fonts at U+F020 to U+F0FF: though I cannot say that I know that other use will cause problems, I think it best to avoid it for other uses, just in case.

I would also advise a little against using hexadecimal codes ending in 00 as I sometime ago noticed problems when keying some of them using Alt codes in WordPad on a Windows 98 PC: a press of the space bar being needed as well. I think that this might possibly have been something to do with an old Microsoft way for inputting some Japanese characters, but I am not sure.

William Overington

30 August 2007
cxhris
Posts: 5
Joined: Sat May 12, 2007 11:54 am
Location: uk
Contact:

Post by cxhris »

thanks for that information. i shall try to digest it.
It will confuse users who use your fonts with Coptic characters
i was going to say 'too bad for the Copts', but perhaps they've had enough to put up with.
...noticed problems when keying some of them using Alt codes in WordPad on a Windows 98 PC

i'm still on windows 98.... my slight concern was that bits of code, or component routines, get reused between applications, and maybe my old computer browser's dislike of certain numbers might reoccur in an up to date application.
....very excellent thing by the way, Font Creator....
William
Top Typographer
Top Typographer
Posts: 2038
Joined: Tue Sep 14, 2004 6:41 pm
Location: Worcestershire, England
Contact:

Post by William »

Regarding Private Use Area code point allocations for special glyphs which might be useful in script fonts, the sixth and seventh posts in the following thread hopefully might be of interest.

viewtopic.php?t=2204

William Overington

9 April 2008
Post Reply