Font Forum

Hi
I just moved from FC8 Standard to FC12 Pro and the AutoKern no longer has the option to ignore positive kerning values when running the wizard. I ended up having to do it using character replacement after copying the OpenType code to Excel.

Please bring back the option to ignore positive kerning values when running the wizard. In FontForge this is called "only move pairs closer together".

Alternatively, the following tools would assist in achieving the same goal:
- sorting by Xadvance
- Selecting multiple kern pairs for deletion

Erwin Denissen wrote: ↑Mon Jul 01, 2019 1:04 pm We've removed that option, as too many people start believing only negative kerning is good. If your font only requires negative kerning, it is likely the bearings are not correct.

I agree. If a font needs positive kerning then it should not be removed. Kerning lookup tables can be trimmed to remove values that are too small to be worth including. For a font with 2048 funits/em kerning values below ±21 funits are not very significant. Trimming small values may reduce the number of pairs by 30-40%

Bhikkhu Pesala wrote: ↑Mon Jul 01, 2019 6:07 pm

Erwin Denissen wrote: ↑Mon Jul 01, 2019 1:04 pm We've removed that option, as too many people start believing only negative kerning is good. If your font only requires negative kerning, it is likely the bearings are not correct.

I agree. If a font needs positive kerning then it should not be removed. Kerning lookup tables can be trimmed to remove values that are too small to be worth including. For a font with 2048 funits/em kerning values below ±21 funits are not very significant. Trimming small values may reduce the number of pairs by 30-40%

Kern 15 funits.png

Shouldn't we have the right to make that decision ourselves? Especially if we are paying money with the expectation that the tools we used in previous versions will also be included in the versions we receive after we upgrade - the purchases are being made under the assumption that we are buying the same tools, as a minimum.

In my opinion, if the tools encourage users to make poor quality fonts they should be improved.

Positive kerning pairs are required to avoid ugly clashes. Automatic kerning pair generation will, inevitably, create some kerning pairs that need manual adjustment. Users can remove or edit pairs that are clearly not correct, but to strip all positive kerning pairs just because one thinks that kerning means only negative kerning is just wrong as it is based on a misunderstanding of good typography.

This is an example from Calibri Regular, a Professional font with thousands of kerning pairs.

Positive Kerning
Without Kerning

Specifically I'm having issues where FontCreator is taking perfectly good kerning pairs with negative values (for example AW and AV) and giving them positive kerning values. Has nothing to do with clashes.

outofstepfontco wrote: ↑Thu Jul 11, 2019 2:23 pm Specifically I'm having issues where FontCreator is taking perfectly good kerning pairs with negative values (for example AW and AV) and giving them positive kerning values. Has nothing to do with clashes.

For such cases, FontCreator would not generate negative values either if there was an option to exclude positive values. As noted above, there may be some cases where manual adjustment is required after auto-generation. The automated process depends on the shape of the glyphs and the white space between them.

What value are you using for the Glyph Spacing with auto-generation?

Take a look at Adding Subtables for Kerning Lookups, which makes it quicker to find and adjust manually those pairs that need it.

outofstepfontco wrote: ↑Thu Jul 11, 2019 2:23 pm Specifically I'm having issues where FontCreator is taking perfectly good kerning pairs with negative values (for example AW and AV) and giving them positive kerning values. Has nothing to do with clashes.

If I'm happy with my letter spacing (no clashes), and FontCreator can be set to block positive kerning, then all positive kerning values will be set to zero, where they can then be eliminated by setting the Trim tool to 1. But without being able to block positive kerning, there's no way to trim positive kerning pairs because kern values can be any value.

It's just a very valuable tool, with a lot of applications. Just bring it back.

We'll consider adding it back if more people like it, but let's wait and see if more people actually comment here.

Erwin Denissen wrote: ↑Thu Jul 11, 2019 6:11 pm We'll consider adding it back if more people like it, but let's wait and see if more people actually comment here.

People post here once every few months. "More people commenting" is a poor measure of demand.