liga vs. calt

paul d hunt's picture

jumping off from the discussion on How Many Ligatures?, i have my own question. are precomposed ligatures preferable to using contextual glyphs that can combine with other letters the way that a ligature would? for example: would it be better to have an f_f ligature, or a contextual first f that would change to a linking version when followed by another f. we already do this for cursive styles, why not for roman?
one positive i can see from this scheme is that you can keep the glyph count down if you have a lot of character combinations that need to be solved with contextual forms, especially if these involve characters that may have diacritical forms. one drawback is that throwing diacritical forms into the mix might actually necessitate precomposed ligatures, thus duplicating work in some situations. anyone else have thoughts on this?

SparkyType's picture

Sorry, I posted in the other thread before reading your topic.

Being able to do multiple glyph subs in a single lookup would be the ideal. The alternative is a bit of a scripting mind bender, especially if you are wanting to sub more than 2 glyphs at a go.

I remember Thomas Phinney promised this feature a few years ago at a TypeCon. Just kidding - he didn't. But seriously, Adobe - do it.

- David

paul d hunt's picture

i don't think you would really need many to many lookups, but some simple chaining contextual substitutions would do the trick. this might become a bit more complex with diacritics, especially if you are using a 'mark' feature, but i think this should be doable easily enough?

dezcom's picture

Them many for one thangs would be much appreciated.


Nick Shinn's picture

If possible, I design f to "ligature" with b, f, h, i, k and l.
Very necessary for news faces, where some "pagination" (layout) applications often don't support even fi and fl ligatures.

John Hudson's picture

I find myself using a combination of approaches, but philosophically at least I favour the contextual glyph option over ligatures. I think ligatures are essentially an artifact of metal type, and probably not veru helpful moving forward. There are exceptions, of course: where ligation of two or more characters results in a completely new, merged shape, as sometimes happens in traditional Greek or Devanagari ligation, then ligature glyphs are necessary. But if the component characters are essentially maintaining their independent identity and only linking together in some way, then I think contextual substitution makes so much more sense. It took a few years of dealing with Arabic to confirm this view, and I think it is only because the typical Latin ligature set is so small that it has not been more of an issue.

By the way, Jelle Bosma used the contextual alternate approach in Cambria, but associated the lookups with the 'liga' lookup, rather than with 'calt'. I queried this at the time, but in retrospect I think it was the right choice.

paul d hunt's picture

thanks John. my current thinking follows yours, i would say. i'm leaning toward the opinion that contextual forms should be used wherever they can be and that true "ligatures" (perhaps more properly "logotypes"?) should be a last resort in instances, as you specify, the merged shapes morph so much as to make a simple contextual alternate form more difficult to work with than than it's worth.

interesting the insight on Bosma's work. i'll have to think on that a bit more. thanks again for sharing your experience.

John Hudson's picture

PS. re. Devanagari and some other Indic scripts. There is a secondary usefulness for ligature glyphs in Devanagari, Bengali, Gujarati, etc., which is as triggers for width variants of the prepostional short ikar (vowel sign), which in traditional writing reaches over the full width of a ligated conjunct.

Syndicate content Syndicate content