in-Glyphs kerning not matching output

[aaronbell]

While reviewing kerning in a recent project, I noticed that the kerning shown by Kern On present in the Glyphs window does not match the ultimate kerning generated by Kern On.

I'm aware that there are cases where a particular pair may not get kerning if it is too far down the frequency chart, but this is not that.

In the attached images, the o-cy and em-cy are shown in the KernOn window set to auto with a kern applied of -21 for the Display Black and -19 for the Display. However, after running KernOn, the output is -85 for the Display Black and -55 for the Display (do note I've rounded these values to the nearest 5 so they're not the exact numbers).

The only way I've found to override this issue is to set the pair as "independent", so KernOn effectively ignores it. Strangely, this also seems primarily problematic on the Display master more than the Text master.

Why would KernOn generate different kerning when previewing in Glyphs vs after running?

[Tim Ahrens]

I assume what is happening is what you also suspect: The pair is too infrequent in the list. This is why it does not end up with the ideal value (as it is generated on-the-fly while Kern On is running). See my explanation in the other thread for more details.

As an experiment, you can increase the value for the pair in question in the frequency list, which you can find by right-clicking the plug-in file (KernOnGlyphs3.glyphsPlugin), Show Package Contents, then it’s Contents/Resources/pair_frequencies.txt. Don’t set it to more than 36000, though.

I’ll be happy to have a look at the file, maybe I can figure out what is going on.

[aaronbell]

Hm. I understood what you wrote in the other thread, but I don't think that's the same as this issue. Here, the kern pair is being touched, it is just resulting in an odd and unexpected output.

That said, I have since updated Glyphs and KernOn and am no longer seeing the issue. Not sure if there was some odd mixture of behaviors that caused it. If I can repro the problem again, then I'll follow up.