They should be easy to set up with the Reinterpolate function.
The diagram I drew is for expanding two brace layers on the edges to five, or one on the middle of the design space to five. What counts is that you set up quadrants. So if you have one brace layer on one edge of the design space set up one on the opposite edge, and you’re done.
@mekkablue ok I think I understand. Still a bug because my solution is not perfect (had to round 55.33 to 55).
Basically:
Was easier to change the width than modify all the brace layers… 55 110 165
My thinking: if width axis has 3 values equidistant, take 188 stem width (black master) - 22 (light master) / 3 = 55.33.
Nevermind, that did not fix it, just made it closer when I had forgotton to change the “Axis Location.” 
There are many brace layers on many weights so for now I may just give up until there’s a fix…
The diagram I drew is for expanding two brace layers on the edges to five, or one on the middle of the design space to five. What counts is that you set up quadrants. So if you have one brace layer on one edge of the design space set up one on the opposite edge, and you’re done.
Somehow missed this in my previous two replies, sorry – will look into it…
@mekkablue still not sure if I understand. tried this, and the character is broken…
edit: if you have a concrete example (this is for /a), I could then extrapolate what to do with the rest of the glyphs… sorry, just having a hard time understanding! a lot of this terminology is new to me
No, in this case you do need extra brace layers because you already have two quadrant. And {22,500} already is a master:
Thank you for your patience on this!
I thought I understood, but I am still getting a bad /a (for example) with this setup:
@mekkablue Just to clarify on how I came to those brace layers, to make sure I understood the designspace you were describing.
Maybe it is a bug, maybe it is related to lack of removeoverlap somehow affecting it (see my other thread), or maybe every single glyph in the font needs to work this way before any one can? I don’t know.
Can you send me your current file again?
Ok, just sent it to support@
/a is the glyph with those layers
I think I got it to work as expected by removing some superfluous/misplaced brace layers. Look at a and n.
The only other thing I did was modifying the AxisLocation parameters to always go from 0-1000 on both axes.
I think this may be a system difference then; still no luck for me on 10.12.1, with latest Chrome Canary with Experimental Web Platform Features enabled in chrome://flags
I suspect you are on 10.12.5 or 13 beta?
What about the files I sent you?
My Glyphs version is 1038.
After updating to 10.12.5 (or maybe through that process of restarting or clearing some cache) /a/n do work. Confusing since I didn’t think that was system level but rather the browser.
/r is still different however. Other letters too. Compare /a/r/n if the width is set very low it is especially noticible.
The r had a brace layer at {100,500}. It works for me after inserting and reinterpolating a brace layer at {100,300}, as discussed above.
Perhaps you want to reread the post about quadrants?
Right, of course. Thank you again for your help on this. I hope a future version of Glyphs does not need the extra layers, but I think I will write a script to create them in the mean time.
You get different interpolation zones, and different vectors.
this is not a very flexible script and poorly coded but gets it done, wrote it ages ago but forgot to share here – https://github.com/jpt/barlow/blob/master/tools/Brace%20Layer%20VF%20Designspace%20Fix.py
It would be a nice feature if Glyphs included support for a parameter to do this
