Too curvey and problematic slurs

• Jun 6, 2019 - 07:59

Here are three examples of slurs that need manual intervention to improve positioning and shape:
Screenshot from 2019-06-06 07-31-29.png
Bars 1 and 5 are as ms 3.1 would create the slurs. Bars 2 and 6 are my manual interventions to achieve something visually more pleasing (to me) with autoplace still active. Bars 3 and 7 are the same as 2 and 6 without autoplace active.

The problems I see are these:
bar1 1st slur: left hand end curves under the slur. I don't see any need to pin the end of this type of slur to the proximity of the note head. Many printed editions don't.

bar1 2nd slur: the parabolic bend is ugly. Again, visually much better if the left hand handle is raised, however, with autoplace active, layout alterations elsewhere will undo any manual corrections, as has happened here. in bar 3 I had turned off autoplace, and achieved a similar shape to bar 2 but bar 2 got messed up again when I added bar 4.

bar1 3rd slur: ms doesn't seem to be consistent in the way it combines the end handle of a slur with a tie. I much prefer to see a slight separation as in bar 2 and 3.

bar5 shows similar problems, but exacerbated by the presence of accidentals. I don't understand what the 1st slur is placed through sharp, but the 3rd is raised a considerable distance above the flat. For me, preferable alternatives are in bars 6 and 7.

I wouldn't say that autoplace is to blame, but in the end, turning it off for all slurs leads to much less work in the final stages of preparing a print-ready score. Autoplace does seem to cause excessively parabolic slurs as in bar 1 bear 2 and bar 4 beat 2. Also, one can't raise or lower an end handle without the curvature of the slur flexing in an unpredictable way or pivoting about a centre handle. With autoplace off, the slur is a more aesthetically pleasing shape; raising or lowering an end handle doesn't cause the shape to flex. I wonder whether there are ways to control slurs from the style settings that would required far fewer interventions?

MS file of these examples: slurs.mscz


Comments

Thanks for the examples! These things tend to be very subjective, so finding a one-size-fits-all algorithm is probably not possible. For the record, originally autoplace would not try to curve those slurs as shown, and the result would be flatter slurs with ends further from the notes. This yielded a number of complaints, so I implemented a scheme that would try to detect such cases and curve the slur a bit more in such cases. See for example #280466: slurs too high, outside staff and #285173: Slurs too flat in some cases. It the fact that we take multiple passes o try to get a curve that allows the endpoints to be close enough that is responsible for the fact that things can change while you're editing - your edit might eliminate the need for the second pass.

It would be possible indeed to have a style setting and/or property in the Inspector to control whether we do the extra pass to curve the slur more.

In reply to by Marc Sabatella

I agree that subjectivity is a major part of this and therefore one can't be overly prescriptive. The templates suggestion is interesting. I'd like to hear how that could be developed. I like the idea of a few more controls in inspector. Its a difficult issue to suite all users all of the time. In the mean time I'll make a judgement on how I tackle slurs on a case by case basis. For me it may be the one place autoplace has to be turned off.

In reply to by Marc Sabatella

This is "subjective" but there is nevertheless a long tradition that can guide what is the appropriate position of slurs. The original post exemplifies an aspect where MuseScore doesn't do a good job.

It surprises me that this issue https://musescore.org/en/node/285173 was seen as a bug, when it clearly isn't (at least not with the example provided in the issue). MuseScore was doing the expected slurs fine. I would mark the "fix" as a regression, since something that wasn't broken got fixed. It can also be that there are other examples that were indeed bad, and that got fixed in that issue, but the example presented there shouldn't be considered a bug. I suspect "fixing" that causes the first weird shape of the slur of the example of the OP (between g4 - c5), which I have never seen in any published edition. For comparison, here is what Lilypond 2.20 does with the mentioned "bug":

slurs fixed.png

As comparison, again, here is what Lilypond does with the example of the OP

slurs lilypond.png

I agree with "bar1 1st slur: left hand end curves under the slur. I don't see any need to pin the end of this type of slur to the proximity of the note head. Many printed editions don't". This is indeed the case in printed editions provided that two adjacent notes have a large interval. However "bar5 shows similar problems, but exacerbated by the presence of accidentals. I don't understand what the 1st slur is placed through sharp, but the 3rd is raised a considerable distance above the flat. For me, preferable alternatives are in bars 6 and 7". The alternative in bar 7 of avoiding the sharp but not the flat is somewhat inconsistent. Lilypond rightly crosses both as shown above (note the slur crossing the sharp in line and not in space). A prompt fix for this should be given high priority, in my opinion.

Regards,
Martín.

In reply to by m.r-botero

I would also add: "bar1 3rd slur: ms doesn't seem to be consistent in the way it combines the end handle of a slur with a tie. I much prefer to see a slight separation as in bar 2 and 3".
This seems to come from the fact that MuseScore doesn't tie the note using the nearest space (compare with the 2nd Lilypond example), but instead puts the tie very far away from the note head, in both cases, outside of the staff. There could be other factors involved as well.

In reply to by m.r-botero

Hello! Yes, slurs are indeed quite subjective. There was quite a bit of discussion and others agreeing it was best to curve the slurs more, that's why we implemented the change, but indeed, sometimes it isn't what you want either. You can disable autoplace to get the flatter slurs that you can then position and possibly bend more yourself. But also, the algorithm uses a couple of different parameters that we could probably expose via style settings so you have a choice of how hard we try to curve the slur and how far from the notehead we allow the slur to be before trying.

I know our engraving expert is looking at writing up some new rules to follow for MuseScore 4, so hopefully we'll be implement many of his suggestions. Of course, again, it's all subjective, and it's still quite possible you might find cases where you don't like the defaults. But hopefully it will be more to your liking overall, and with more control offered over the defaults.

In reply to by Marc Sabatella

"Hello! Yes, slurs are indeed quite subjective. There was quite a bit of discussion and others agreeing it was best to curve the slurs more, that's why we implemented the change, but indeed, sometimes it isn't what you want either".

I would be interested in seeing the examples put in that discussion, but I suppose it's not that relevant anymore ;-).

For what it's worth, here's an excerpt of the Händel Trio Sonata Op. 2 No. 6, Edition Peters, m. 14:
IMG_0587.jpg

Here's what Lilypond does:
händel Lilypond.png

And here's what Musescore does (it has my custom style which I don't know how to "deactivate". Slur positioning is, however, Musescore's default):
händel Musescore.png

Musescore's first slur is indeed too close to the first note head.

In reply to by m.r-botero

In general, rather than try to guess at rules by looking at any one particular edition or what any one particular other program might do by default, we try to refer to established texts that lay out specific rules we can follow. And unless there is a preponderance of evidence that she is an outlier, or normal go-to is Elaine Gould's "Behind Bars". In the examples she provides that are closest to this one, she shows the slur starting half a staff space from the notehead, which is what she lists as the minimum distance. Which is therefore what we do also.

We do recognize Gould is not the only authority, but doing what she recommends is seldom a terrible idea in cases where things are subjective. In any case, this is area where we will likely be getting additional guidelines from our own engraving expert.

To me, the LilyPond example is quite far from anything Gould shows or recommends. The Peters example is closer.

In reply to by Marc Sabatella

Sure, Gould's rule is fine as a general rule, but it certainly doesn't work for large intervals. Finding printed examples for a given rule is finding the "evidence" for making such a rule. That's why I would like to see examples of that rule applied to large intervals. If you happen to remember published examples of the old discussion, I'd be happy to take a look. Musescore's behavior is new to me. I can't recall having seen something like that ever.

In reply to by Marc Sabatella

The issue https://musescore.org/en/node/285173 only links to this thread here in the forum so finding the corresponding discussion is quite difficult. I could however find the Telegram discussion. Unfortunately there are no musical examples but instead only a discussion about Bézier curves. I understand now a bit of how the algorithm works, and it makes me also realize that in traditional engraving the curvature of slurs has indeed a limit. In the only example provided in the discussion of the algorithm at work, it seems to me that a maximum of three iterations is simply too much. Two would be enough, in my opinion. I hope @oktophonie can agree with me on that, at least if he doesn't want to re-write that part of the code from scratch with a different type of curve/behavior. I also hope I never have to compile Musescore myself only to be able to use a more sensible configuration of slurs, provided that there is an "easy" way for me to change that 3 to 2 somewhere in the code :-).

Do you still have an unanswered question? Please log in first to post your question.