ID doesn't have the same kerning dictionary functionality that was available in some other layout applications. One workaround is to create a GREP style that finds the pair, then applies a tracking value.
I've thought about the GREP/tracking solution, but have dismissed it as setting tracking along two characters affects not just the two selected characters but also the space between the second character and the third unselected character.
So a GREP style would have to find the two-character pair, but then apply tracking to just the original first character.
I'm not sure how to do that.
Also, you'd have to create a GREP style with different tracking values for every point size of the text and every font style.
And then repeat the process for every character set you want to change.
It would be a nightmare to keep track (sic) of all those setting
I'm not dismissing your suggestion. I sincerely would like to use the idea, but have never figured out how. Could you elaborate on what you would do to keep it simple?
One way woud be to find the first character with a positive look-ahead for the second. That's really not any harder than finding the pair.
And I don't understand why you think you need to do this for every size. Tracking values are in ems, so they're relative. Not to mention that a GREP style is applied inside the paragraph style, so it's being applied discriminately to text that you know the face and size for, so you really only need to set up a group of character styles with different tracking values, then apply the appropriate one for the particular case. You might get by with as few as three or four, or you might need forty, depending on how many pairs you need to deal with, and how diverse the amount you need to kern them. Frankly, if there's more than two or three special cases that need adjustment, I'd be looking for a better font.
Is it not more productive to change the kerning directly into the font? I did it a long time ago about the Gill Sans font modifying it with Fontographer!
The problem with "fixing" your font is that it causes problems in a collaborative workflow where someone else doesn't have your "fixed" version of the font installed, so the kerning is different and text reflows. This was even more of a problem in the days of sending native files to a printer.
Uwe, one can accomplish the same effect using negative Tracking in a character style, though.
In general I agree I would rather avoid using a character style—whether via tracking or kerning—and just use another font if it bugged me.
The pair of "Ke" highlights the issue of kerning I would have made tighter in most fonts I use regularly for headlines. But on text smaller than a headline, I think the optical kerning ID provides automatically is more readable. Which is one reason I have never made kerning tables in QXP since decent fonts have been available (really, not since Type 1 fonts).
I totally agree with you. It is a big problem. That is why the name of my modified font becomes "xxx_OWK", "xxx" for the original name of the font and "OWK" for Obi-wan Kenobi", the person who modifies the font.