I'm working with a collaborator on a font which includes Latin and Cyrillic. It is a "marker" font, so it doesn't have super-regular glyph shapes, making kerning that much harder to do – and that much more desirable to automate!
However, we have run into a few problems, and I'm wondering how much of it can be solved with better understanding/usage of Kern On, vs how much just requires manual kerning – and that what the interaction is like between manual and KO kerning.
And early problem: after I did a pass of Latin kerning with Kern On, making a collection of models, my collaborator reported that the `ДД` (/Decyr /Decyr) pair was really spaced out ... and indeed, it had +102 and +110 kerning, setting it way too far apart, especially in this font, in which we are using a handwritten, triangular form of this character.
What's more, while KO was open, I found that I could not even adjust this kern downwards by more then about 10 units, when we needed to bring it down by around 50 units. (At least, the slider wouldn't allow me to do so, and KO was fighting my keyboard inputs for kerning the pair.)
I finally got the kern I wanted (around +60) by closing KO and entering it manually, directly in Glyphs. Then, when I opened KO again, it seemed to accept the pair as a model.
This brings me a few questions:
1. Is this perhaps a thing that could be improved in KO? I can imaging a logic for adding so much space, to keep a certain minimum distance, but the minimum distance is all below the baseline in this case, so it should be weighted a lot less than what happens between the baseline and capheight.
2. If my collaborator were to go through and fix kerns manually (after proofing to find them), and then I Git merged their file into mine, would KernOn detect those manual kerns as new models? Or, would it throw them out next time I calculated kerning?
Thanks for any insights!
1 post • Page 1 of 1
- bold.png (173.88 KiB) Viewed 393 times
- light.png (176.7 KiB) Viewed 393 times