Page 1 of 1

[FIXED] Kerning Values Different for Identical Glyphs

Posted: Sat Dec 17, 2016 8:32 pm
by Bhikkhu Pesala
I have two glyph classes — r_lc and r_lc_top. Apart from the accents above the glyphs, they are identical in terms of bearings and advance width, so why are the kerning values returned by Autokern so different?
r_lc.png
r_lc.png (9.49 KiB) Viewed 5674 times
r_lc_top.png
r_lc_top.png (11.3 KiB) Viewed 5674 times

Re: Kerning Values Different for Identical Glyphs

Posted: Sun Dec 18, 2016 2:27 am
by MikeW
racute hyphen is exactly the same value for me (-53 x-advance) in your font project.

But unless there is a reason not to, I would combine both of these classes anyway.

Re: Kerning Values Different for Identical Glyphs

Posted: Sun Dec 18, 2016 7:00 am
by Bhikkhu Pesala
MikeW wrote:racute hyphen is exactly the same value for me (-53 x-advance) in your font project.
Thanks for looking. I adjusted the values manually after taking the screen shot, but the calculated XAdvance value for r_lc_top is -8, as shown in my screen shot.
MikeW wrote:But unless there is a reason not to, I would combine both of these classes anyway.
Yes, there is a reason for separating the “r”s into two classes. I use the naming convention “@class_name_top” for those glyphs that are likely to clash with the top of capital T, V, W, Y, etc.

I am still refining the kerning classes to have as few classes as necessary, while avoiding overlaps. The method of “@class_name_first” and “@class_name_second” works better in some cases. I have attached an updated project file to this post and removed the older version from the previous post.
Tr.png
Tr.png (24.78 KiB) Viewed 5662 times
Tr_top.png
Tr_top.png (27.45 KiB) Viewed 5662 times

Re: Kerning Values Different for Identical Glyphs

Posted: Sun Dec 18, 2016 10:38 am
by Erwin Denissen
MikeW wrote:racute hyphen is exactly the same value for me (-53 x-advance) in your font project.
I've just run Autokern (right-click the lookup) with glyph spacing factor 27, and both get 61.

Re: Kerning Values Different for Identical Glyphs

Posted: Sun Dec 18, 2016 11:03 am
by Bhikkhu Pesala
Erwin Denissen wrote:
MikeW wrote:racute hyphen is exactly the same value for me (-53 x-advance) in your font project.
I've just run Autokern (right-click the lookup) with glyph spacing factor 27, and both get 61.
I tried with spacing value of 27 and got positive kerning values of 39 and 77.

With my spacing value of 23, I still get -53 and -8.

I noted that c_lc hyphen and c hyphen give the same results (-104), but z_lc hyphen (-88) and z_lc_top hyphen (-83) do not.

Re: Kerning Values Different for Identical Glyphs

Posted: Mon Dec 19, 2016 9:46 am
by Erwin Denissen
I'm puzzled.

I've reduced the pairs to:

lookup PairAdjustmentLatin {
subtable Subtable_0 {
pos @r_lc @hyphen <61> <0>;
pos @r_lc_top @hyphen <61> <0>;
}
}

And right-click the lookup to Autokern keeps them the same.

Re: Kerning Values Different for Identical Glyphs

Posted: Mon Dec 19, 2016 10:04 am
by Bhikkhu Pesala
Erwin Denissen wrote:I'm puzzled.
Me too. I tried with just those two pairs and still get the same results as before:

53, -8 @ 23 space factor
39, 77 @ 27 space factor

So, at least it's consistently wrong! :?

We're both using the same project file, and the same build 10.1.0.2257 so what else could be different? Is it the rendering dll again?

Re: Kerning Values Different for Identical Glyphs

Posted: Mon Dec 19, 2016 11:28 am
by Erwin Denissen
Bhikkhu Pesala wrote:
Erwin Denissen wrote:I'm puzzled.
Me too. I tried with just those two pairs and still get the same results as before:

53, -8 @ 23 space factor
39, 77 @ 27 space factor

So, at least it's consistently wrong! :?

We're both using the same project file, and the same build 10.1.0.2257 so what else could be different? Is it the rendering dll again?
I was testing with a more recent version; not released yet. Somehow that version works fine, so I might have accidentally fixed the problem.

Re: Kerning Values Different for Identical Glyphs

Posted: Tue Dec 20, 2016 10:40 pm
by Bhikkhu Pesala
The latest update (FC 10.1 build 2272) has fixed this bug.