Font Forum

Having just updated my tutorial on Kerning Classes for FontCreator 10.1, I remembered that the Autokern feature is flawed, and really needs some improvements.

There is already an option not to exclude lowercase-Uppercase pairs like oT, which I think no one ever uses, except perhaps graphic designers who would do it manually anyway to create a logo.

Another option is needed to exclude lowercase-lowercase pairs.

Ideally, I think this feature should use a kerning text file, like FontCreator used to in earlier versions. Designing the kerning pairs text file to suit all languages that use Latin scripts would be difficult, but there are surely some OpenSource files that could be used.

I think Autokern does not cater for Cyrillic or Greek, let alone other scripts, but text files could do this more easily than the Autokern algorithm.

I can see why one would exclude upper-upper, but why lower-lower?

Erwin Denissen wrote:I can see why one would exclude upper-upper, but why lower-lower?

I see no reason at all to exclude Upper-Upper. Pairs like AV, AT, TA, LT are those needing the greatest values. However, there might be a case for not including positive kerning pairs for Upper-Upper like AA OO etc.

Most lower-lower are not needed. One can set the lower limit to 20 funits to exclude some of them, but it's far easier to add them manually for the few that do need it like y, x-, and maybe ko.

Running Autokern while updating my Kerning Classes tutorial with a minimum kern value of 10 funits, resulted in 275 kerning classes, and a whole ship load of lower-lower pairs I would never bother with, and which would never occur in normal text.

Here's the kern lookup table from a new font after running the Eastern Europe transform script with a minimum kern value of 10 funits and another with a minimum of 20.