Le 09/03/2022 à 17:13, Matthew Fong a écrit :
Dear Jean and LilyPond friends,
My apologies for not initially including this, and it took time to get
this reduction; At the same time, producing this example code raised
additional observations about how the slur overrides work with other
variables, which I can use some insight about. Thank you all in advance!
Please note in the code, I have marked these observations by a comment
starting with: % [Interaction: ... ]. Commenting out noted lines out
one at a time changes the Slur override behavior, and it seems rather
unexpected.
File is attached (/_reduction.ly <http://reduction.ly>/), containing
all the functions and the quilismaPathGreciliae PostScript path.
*Observations for Slur above first note group without quilisma*
1/ Changing the staff size has an effect on Slur overrides (Slur too
close to notes)
2/ Altering the LyricHyphen.minimum-distance has an effect on Slur
overrides (Slur too close to notes)
3/ Changing the lyrics has an effect on Slur overrides (Slur too close
to notes)
4/ Interaction between Slur override settings
a/ The desired result (nicer slur)
\override Slur.height-limit = #2.5
\override Slur.details.free-head-distance = #1
b/ Increasing Slur.height-limit results in Slur *too close* to notes
\override Slur.height-limit = #3
\override Slur.details.free-head-distance = #1
c/ Increasing Slur.details.free-head-distance gives the desired
result, similar to (a)
\override Slur.height-limit = #3
\override Slur.details.free-head-distance = #1.5
5/ Finally, there is the original question about why the Slur above
the first note group *with the quilisma* remains unaffected by any
Slur overrides
*Graphical explanation*
/Desired result [image]
/
Slur too close [image]
Many thanks,
Matt
Thanks for the examples. This seems to work better:
\override Slur.height-limit = 10000
\override Slur.details.head-slur-distance-factor = 10000
Honestly, I can't explain why. Slur formatting is a very involved
process. All these parameters only have an indirect influence on
the slur appearance. They factor into demerits ("ugliness scores")
that LilyPond computes for a bunch of possible slurs, to pick the
best one. That's why you are observing all-or-nothing effects: since
there is a staff line and LilyPond thinks the slur would be ugly
if it came too close, depending on the specific values of the parameters,
it will choose one way or the other, to make the slur cross the staff
line or to make it stay below it. Also, when you change horizontal
spacing and such, this modifies the set of configurations that
LilyPond tests in the first place. This is all quite tricky, and
designed to yield the best results possible using heuristics at
the expense of not being very straightforward to tweak. Being no
expert of the particular code, I can't tell you the details of this
art. If you require specific slur shapes, it is best to use \shape
to refine the control points precisely. (In 2.23, you also get
\vshape which provides a visual aid in this process.)
Best,
Jean
