Kerning exceptions and Compress kerning issue

Hi Team

The Compress option in the kerning table seems to be behaving a little unexpectedly and causing me some issues.

I have i (dotless) that shares a left have kerning group with n.
All my i accents share the same n kerning group
My f/n pair has no kerning between them, but I made several kerning exception for various f i accent pairs e.g: fí fĭ fî fï fì fī fĩ
On a kerning table Clean up, a value has been added to all my f/n group pairs that before had no kern value (the added value seems to be the lowest kern value of my exceptions).

I would have though that rather adding a kern value where none existed, a zero would be added.
I imaging this will probably cause other kerning issue throughout the font where no value has been previously added but exceptions exist.

Is my process wrong or is this a bug?

This is exactly what the cleanup is supposed to be doing. Copy values from glyph pairs to the class pair. And this is one of the obvious fails. But why do you run the compress command at that stage anyway. That is meant to be used early or when importing old data.

1 Like

Ah, Thanks Georg.
I had assumed the command would run some magic algorithm to reduce the number of values while somehow keeping everything intact.
I’ll leave this button alone for now!

Still not quite sure I understand what you want to achieve.

But perhaps simply delete the exceptions (filter kerning with the magnifying lens in the Kerning window, then select the pairs and delete), and add the exceptions you need with the Auto Bumper. This is one of the mekkablue scripts in the Kerning submenu. Look around there and see if there is something else that may help you. The Readme explains every script, and there are tooltips in the Script menu and the script UIs.

I know it is not the most exciting topic, :sob: yet I cannot spare you from digging in a bit if you want to take more control of the kerning.

1 Like

Thanks @mekkablue Don’t worry about it not being exciting. I seem to spend spend so much time kerning that I’m happy about digging in to help simplify it!

While I’m careful to ensure my kerning isn’t compensating for poor spacing, and that I’m not over kerning, each weight of almost 500 glyphs still contain almost 2000 kerned pairs. I’ve grouped them where possible, deleted small pairs but maybe this ratio is about right for a san serif?

1 Like

That seems like a lot.Half of that amount would be healthy, I guess.

1 Like