kerning: all glyph combinations in a font

Martin LAllier's picture

Maybe my question has been asked somewhere else already, but search not beign available yet...

Is there a way, either from Fontlab or from a script (maybe robofog?) to have a list of all the glyph combinations possible. Lets say an output to a txt file.

For instance, I'm currently working on a 620 glyphs font and would like to quickly check out all the possible miss matches in the metrics and kerning. I've used the KernKing by Mr Cabarga but I've yet many more glyphs that are not covered in there.

Thanks in advance!

.'s picture

There's a little thing called an AFM file...

eomine's picture

... And another one called class-kerning.

It's not a good idea to check every possible combination of letters. With 620 glyphs you have more than 380000 possible pairs; it's just too much to handle.

Martin LAllier's picture

My goal is just to identify problematic pairs.

The best recommendation I've got so far was on typographe.com/forum, by JF Porchez, “do a 'listing manually' of all the pairs (aba, aca, ...) in the font, then check them all”.

Longer but more rigorous. Just wish this long list could be generated automatically... :)

dimitre's picture

Hello, I was designing a font that I needed to identify some common pairs too, and I developed this application to make statistics based on some online sample text

http://dmtr.org/experimental/pairs.php

I used it to define priorities in kerning and legatures of my font, based in some TXT books in spanish, portuguese and english

dezcom's picture

dmitre,

Can you explain what you have done?

ChrisL

grod's picture

yeah, it looks interesting but it isn't completely self-explanatory.

dimitre's picture

ow sorry, here it goes:
it is a software to analyze the character pairs occourence
for example, I can take a txt book
http://library.beau.org/gutenberg/1/0/0/0/10001/10001.txt
and discover the frequency of pairs.
the funny thing is the pair of characters depends of language you use, of course, and of writing style too

here you can see some examples:
http://dmtr.org/type/FluidTypeface/stats.php

elliot100's picture

"the funny thing is the pair of characters depends of language you use"

This was recently brought home to me when I tried to use some common fonts to set some Albanian. Seems "gj" isn't a common pair in other languages...

hrant's picture

Logjam. It's on the freeway about half of every day.

hhp

elliot100's picture

good point. what about ej?

[edit: oh yeah, bluejay. Well whatever the problem pair was, the result was spaced like Shane McGowan's teeth]

hrant's picture

Reject.

My secret: chapter 3 in Ross Eckler's "Making the Alphabet Dance".

hhp

Martin LAllier's picture

Another exercise in pairs visualization - http://www.m-i-b.com.ar/mib/letter_pairs/eng/

hrant's picture

Now that's some cool visualization - bravo.

hhp

sim's picture

here you can see some examples:
http://dmtr.org/type/FluidTypeface/stats.php

This link does'nt work for me.

dezcom's picture

The biggest kerning pair needs are either all caps or cap to lower case-- I should also mention puctuation marks.
Are there frequency stats on these? The usual T, Y, V, A, ones are obvious but I have not seen real stats.

ChrisL

dezcom's picture

I try to shift sidebearings on caps so that the right is set to sit well with lower case and the left sidebearing is larger for all caps setting. This reduces the number of kern pairs but there are still pairs needed.

ChrisL

Syndicate content Syndicate content