My character set includes both Ä and Ǟ. In the latter, I want to lower the dieresis slightly as to not make the double accented letter too tall. I did this using two anchors: The A has a “top” anchor on the cap height, and a “stacked” anchor slightly below. The dieresis has both anchors ("_top" and “_stacked”) at the cap height.
The solution is a bit cumbersome, because
a) for some reason, Glyphs prefers the “stacked” anchor over the “top” anchor, and I have to reset the Ä to use the “top” anchor position.
b) This also means the accents with “_stacked” anchors gets placed lower when I use combining accents. I can fix this by removing the references to “stacked” and “_stacked” in the mark feature
c) the results differ in precomposed and decomposed setting.
To my questions:
• What determines which anchor to prefer when there are more than one?
• Is there a better way to shift the y position of the accent closest to the base glyph down when there are more than one accent? E.g.: Ä with the dieresis positioned by the “top” anchor, Ǟ with the dieresis positioned 50 units lower.
For compounds: You can use top and top_stacked anchor (note the suffix starts with an underscore) in the base glyph. Then, in the compound, you can click on the component and, in the grey info box, switch to the alternate base anchor.
For mark positioning: You can also employ (a) an alternate
dieresiscomb.stacked and put it in the
calt feature, or (b) add something like this to
pos dieresiscomb' <0 0 -50 -50> macroncomb;
Keep in mind you cannot preview GPOS features within Glyphs.
- There must be corresponding xxx and _xxx anchors in base and accent.
- And then, it is alphabetic, so aaa will come before bbb.
- Lastly, by default, base anchors (top) before underscore suffix variants (top_alt).
Thanks, that makes sense.
note the in the base glyph
A word appears to be missing in this line.
I wonder if I’m misunderstanding the tutorials. The mark to mark articles says that the features should be built automatically if I include “_top” and “top” (“top_center” I presume also works) anchors in the combining accents. But it isn’t. Is there something else I have to do?
Thank you, fixed it.
They do. But because the numbers are interpolated for every instance, you cannot see the feature in the Features tab of the Font Info. If you want to see the feature code, take a peek into the Temp folder. Or just try it in InDesign.
I know that, but my generated font does not have an mkmk feature. I also studied the contents of the temp folder.
mkmk automation, the anchor names must be matching: So,
_topcenter. Or even
So in your combining accents, you could have
mark, and put in
_topmkmk slightly above that and
topmkmk above, see the image below. That way, you can have different anchors for
mkmk. That would be a solution for your problem with out any manual feature coding.
EDIT: Halt your horses, mkmk automation doesn’t work if there are two underscore anchors in it. So maybe the contextual kern thingy may work better after all.
Thanks for helping me figure it out. However, once I got it exported, I noticed your contextual thingy doesn’t really work. The dieresis (what is the plural of dieresis?) should sit at the same height. Precomposed left, decomposed right.
Is this an InDesign screenshot? The problem is that InDesign takes the precomposed adieresis as soon as you type a-dieresiscomb. So you do not see the decomposed version here.
I wrote the
pos line wrong up there. It should read:
pos @COMB_MARKS_lc' <0 -50 0 0> @COMB_MARKS_lc;
pos @COMB_MARKS_lc @COMB_MARKS_lc' <0 -50 0 0>;
And that works pretty well, as long as I take a letter that does not have a precomposed variant that InDesign could default to:
if you add:
sub adieresis by a dieresis comb;
to the calt and activate the Global typesetter in indesign, it works with the
adieresis, too. (you need to add decomposition for all other precomposed glyphs).
Thanks guys. I think there are to many obstacles to get this working reliably. And it not such a big deal anyway. On a side note: Could something like this potentially go in a vkrn feature?
Not sure if macron man approves.