Kerning pairs.

CFCG's picture

I was looking for examples of pairs kerning (for Latin 1). Could somebody tell me where can i find examples similar to the ones published here?

http://www.sudtipos.com.ar/test01.txt
http://www.sudtipos.com.ar/test02.txt
http://www.sudtipos.com.ar/test03.txt

Thanks!


Carlos Fabián Camargo G.

ben_archer's picture

Hi Carlos – this link is already in the wiki here under 'Tools';

http://logofontandlettering.com/kernking.html

You should also look for the Emil Ruder spacing test (because I'm sure someone has it online). I found both of these very useful, but only discovered them after I had typed out all the kerning pairs and all the pangrams I could think of in QX.

charles ellertson's picture

I went and looked at kernking. I suppose it can be useful, but I did notice it left out one of my favorite "pairs," namely Yg. I once had to set a book involving the Artherian legend, and Arther's mom, Ygraine, came up a lot. It should go without saying I had to go back & re-kern the font; I'd missed it in my early effort.

As a typesetter, I've been kerning fonts since 1980 or so. That means that I've had a lot of practice, but it also means I'm a bit set in my ways, so you can take what I say with a grain of salt. At this point, If I'm serious about a text font, I look at all combinations of letters and punctuation, as lowercase, as cap-lowercase, and as cap-cap.

Since the advent of Fontographer, and now FontLab, it has gotten easier. Both have a metrics window that lets you put up whole words, with spaces. I make sure the "metrics window" has about five words, with spaces, and then run through the pairs. With tricky letters, such a cap-V or cap-W, I make sure I look at not pairs, but triplets -- if you use WAVE for kerning the "A", you'll wind up with a very different decision than if you use WARF. And of course, same with the lower case.

Another thing you will notice, especially if you are making your own font, it that you can pick sidebearings for the caps to either work well (minimal kerning needed) with other caps, or work well with lowercase letters, but rarely both. One or the other will needs more work.

Don't forget the numbers, and numbers with endashes, commas, and periods.

You will notice -- I hope you will notice -- that too tight is worse than not-quite-tight-enough. The font itself will give you some hints on how it should be kerned & spaced. I'll change sidebearings & re-kern if need be. The lower-case "f" is a typical candidate. In order to skip kerning with the vowels, some designers will give it a large negative right sidebearing. When the "f" is the last letter in the word, the wordspace will get too small. So they'll add a "f space" positive kern. But that's a patch, if you get it right for "of Wilbur" it will be wrong for "of anything". Better to get the sidebearings right & deal with the letters.

After almost 30 years of practice, It takes me about 4-6 hours to kern a font for the Latin alphabet, and maybe another hour after I see the first book printed, and a little bit of "font maintenance" from time to time as I see the results from other books. That's a plus a compositor has over a type designer, I suppose. But this is a small amount of time to spend on a font; far far less than it took the original designer.

Miguel Sousa's picture

> After almost 30 years of practice, It takes me about 4-6 hours to kern a font for the Latin alphabet

Charles, I'm curious to know what you mean by "Latin alphabet". Does it include all the combinations below?

— lowercase / lowercase
— lowercase / punctuation
— punctuation / lowercase

— uppercase / uppercase
— uppercase / punctuation
— punctuation / uppercase

— smallcap / smallcap
— smallcap / punctuation
— punctuation / smallcap

— uppercase / lowercase
— uppercase / smallcap
— lowercase / uppercase

— numerals / numerals
— numerals / punctuation
— punctuation / numerals

charles ellertson's picture

Pretty much. The only lowercase/uppercase pair I routinely look at is "c", as in Mac- or Mc- (All capitals). Punctuation/uppercase is usually limited to the apostrophe (Unicode 2019), the open & doubleopen quotes, and periods with a few abbreviations such as U.S.A and D.C. where the period needs a kern to the following letter (particularly the "A"). The same would be true of punctuation/smallcap, since abbreviations & acronyms are often set in small caps.

I dunno if I should include small caps in that optomistic time estimate; it would certainly push it toward or more likely over the 6 hour figure. Too, the itals & bolds tend not to have small caps, though I usually make up a rudimentary set [where the EULA allows this, of course ;)]. The occurrence of ital or bold small caps is usually so low in the work I do that they could be handled in the job file -- esp. when a designer has chosen a font no one will ever use again.

We all evolve our routines, and since I primarily set scholarly monographs – and not even the jackets – certain kinds of things tend not to arise. And as I said, I have to go back & fix – and add – things with the kerning from time to time.

Another thing that helps me move along is I only try to get things right for text-size settings, say 8 to 12 point. I've always felt that in display setting, you have to do your kerning as it comes up. What is right for 18-point is not right for 72-point.

Hope that answers the question,

Charles

gabssnake's picture

This is just a way of looking at the problem but, shouldn't we design kern tables for individual languages? Or to take it further, for individual publications in their particular context (country, particular way of writing, most common words, etc)? Like for the special case Charles mentioned above, with Ygraine?

And if writing the tables for particular languages, shouldn't we use sample texts specially designed for those languages?

The "My" pair for instance, would be really rare in Spanish, except for the cases of proper names such as "Myriam" (Which isn't very spanish-like by the way). So it wouldn't be of huge relevance in the tables. But not the same applies for the English language, were the pair "My" is very common.

I think that maybe samples like Kern King work best when getting to know how the font works with the default spacings, but becomes less useful when writing the kern tables. For those, it would be useful to design texts according to the language/country the font will be used in, and ideally offer different packages for different countries.

Does anyone know if theres available data about the most common (irregular) pairs on English or other languages?

Perhaps there is a list of the irregular pairs in the basic Latin alphabet?

Regards,

__________________________________________________
www.gabssnake.4t.com

CFCG's picture

Thanks to all for your valuable help


Carlos Fabián Camargo G.

ben_archer's picture

FWIW, Carlos Fabián you're welcome. I've just spent 15 minutes admiring Alejandro Paul's beautiful work at

http://www.sudtipos.com

and although I don't read Spanish that well, I didn't find any kerning pair test documents on the site, so they have probably revised their content at Sudtipos since you posted your question. The Emil Ruder spacing test I mentioned earlier is, of course, in Jared's response to Hrant at

http://typophile.com/node/10879

CFCG's picture

Thanks!


Carlos Fabián Camargo G.

Syndicate content Syndicate content