opensheetmusicdisplay
opensheetmusicdisplay copied to clipboard
Incorrect tie curvature
I am filing a few bugs referencing this sheet.
This is bug 1 on the OSMD screenshot.
The tie curvature should be pointing up, as per MuseScore.
The <tied>
element in your musicxml file doesn't have a placement or orientation attribute as far as I can see, so the right information to not do automatic tie direction is missing.
(<tie>
is for sound, <tied>
for notation)
It works when you add placement="above"
to the `
Thanks for reviewing this. Please note that as per https://www.w3.org/2021/06/musicxml40/musicxml-reference/elements/tied/, placement
is not required. I assume (because it's not mentioned) that the default is above
.
Happy to submit a PR to default placement
to above
if you will accept it.
I'm not sure that you can assume that the default should be above when the MusicXML spec doesn't mention any default placement/orientation.
(edit: OSMD doesn't default to upwards yet, edited out)
Oh wait, I still had the placement attribute edited in my xml file. My bad.
The upwards default (here only for the highest notes) to me looks overall like a worse result for most of our samples, especially for monophonic songs with lyrics.
Brooks - West Point:
Bach Preludium:
visual_diffs_tie_direction_default_upwards_for_highest_notes.zip
But we could add a DefaultTieDirection EngravingRule. You would set that to upwards? (Maybe it looks better for your typical scores)
You would set that to upwards?
Not sure anymore :cry: I will check music notation books on their recommendations.
I asked on the MusicXML forum and got this answer:
the upper part takes upward-curving ties, the lower part downward-curving ties
How does that sound as a reasonable default?
It sounds reasonable (though it doesn't supply any reasons for doing it that way), but it looks worse for most of our samples. So if anything I'd add an EngravingRule and make downwards the default.
According to BB [Behind Bars, ed.], for single notes there are two cases:
- stems in same direction => tie curves away from stems
- stems in opposite direction => tie curves away from middle line
- stems in same direction => tie curves away from stems
Ah, that applies to (and confirms) our samples where it does look better like this. It also applies to op's sample though. In any case, I'd be fine implementing this standard and/or a default tie direction.
Thanks for the info, Rodrigo!
There is a lot of info in the book related to ties (several noteheads...) I recommend you to look into it, before going in the implementation.
Sure, if necessary, good to know. Though we already happen to do it correctly in our samples, and the current info is sufficient for an improvement, so the rest is currently low priority. Ideally one would aim for 100% accordance with Behind Bars, but there are more important issues right now.