Kerning, GD library of PHP

Hi folks,

I seem to be having some issues regarding kerning, I have attempted to upload a font twice now to dafont, and both times an error is reported, such that the kerning does not display correctly - Dafont uses the GD library of PHP which I presume is affecting how the font is displayed.

I have no more information to hand unfortunately, though I suspect this may be important in other ways. The font also doesn’t kern properly in microsoft word (Mac, 2011). This is despite my reading some articles on the issue and implementing their suggested solutions.

The font has 187 glyphs and 871 kerning pairs.

A shot in the dark, but perhaps someone has had experience with this and could assist?

The issue might have to do what some environments still do not support GPOS kerning but need an oldstyle kern table.

But if you have so many kerning pairs with so little glyphs, your spacing is bad. I suggest to remove all kerning and start over with the spacing. Have a look here: glyphsapp.com/tutorials/spacing You only add kerning for asymmetric glyph combinations like T+o or L+T.

Thanks for the quick reply, the font in question is a script font, so some level of kerning is definitely necessary.

Altering a side bearing for one glyph may result in proper kerning vs one or two others, but thats it. So before I begin kerning I ensure the side bearings are set to give the greatest number of glyphs kerned properly, and then begin with the kerning to complete the rest.

I hope this makes sense, but I don’t know how to avoid GPOS kerning in this instance. Is there a way to use an oldstyle kern in glyphs?

I’m sure even for script fonts you need less kerning.

Ok thanks Georg, turns out I’ve been doing it all wrong.

Font is updated, kerning all sorted with 54 pairs.

Much appreciated! Oh and by the way, Glyphsapp is incredible :slight_smile:

Kerning is off by default in Word. The user needs to turn it on via the Font dialog.

I had attempted to turn it on, to no avail.

Deleting kerning and implementing proper spacing resulted in 54 kerning pairs compared with the original 871.

That seemed to solve the problem.

1 Like