Playback fixes for hairpins in combination with compound dynamics #24569
Conversation
cbjeukendrup
force-pushed
the
hairpins_compound_dynamics
branch
from
214a8dc
to
f2f4280
Compare
|
Tested on MacOS 14, Windows 11, Ubuntu 22.04.3. Approved |
`nominalDynamicLevel` returns `Natural` when there is no dynamic exactly at the requested tick
- Refactor: replace dynamicLevelFromType with more specific functions for specific kinds of dynamics - Make the list of single note dynamics more complete - For hairpins that start with a compound dynamic that's built into the hairpin: use the 'to' dynamic of the compound dynamic as the 'from' dynamic of the hairpin - For hairpins that end with a compound dynamic, same, but the other way around (instead of treating a compound end dynamic as "Natural") - The `spannerTo -= Fraction::eps().ticks();` trick should happen not when the nominal hairpin end dynamic differs from the nominal dynamic after the hairpin, but when the *actual* end dynamic differs. - When the hairpin's own nominal end dynamic is not used because it does not match the 'direction' of the hairpin, this end dynamic was always applied just after the hairpin. But this should not be done when there is already a dynamic at the end tick of the hairpin. That's not because it would override that existing dynamic, because in this case the 'shorten hairpin by 1 tick' trick has been applied to prevent such overriding; instead, it's because it would write this end dynamic 1 tick too early because of that trick, which means that it will apply to exactly one tick, because after that the existing dynamic applies. This is useless, so instead we will discard the end dynamic in this case.
cbjeukendrup
force-pushed
the
hairpins_compound_dynamics
branch
from
f2f4280
to
369a13a
Compare
cbjeukendrup
force-pushed
the
hairpins_compound_dynamics
branch
from
369a13a
to
e334c7b
Compare
|
OS: Linux Mint 20.1, Arch.: x86_64, MuseScore Studio version (64-bit): 4.4.0-242390800, revision: github-musescore-musescore-0fcd11b Does this commit fix the hairpin playback problem in measures 202-205 in the attached file (beginning of variation 8) If not I'll open a new issue. |
|
@rgreen5 Unfortunately, this PR doesn't solve that issue. It looks like the weirdness in that score is caused by the invisible dynamic markings (rather than the hairpins themselves); the dynamic markings are in voice 2, while the hairpins are set to affect all voices. We'll need to think about what the exact logic in such a situation should be. So opening a new issue is indeed the right thing to do. Thanks! |
Resolves: compound dynamics were not (correctly) considered when looking for end dynamic of hairpin
(related to https://musescore.org/en/node/367942)
Seems like that was a 4.3.2 -> 4.4.0 regression
See commit messages for details
Builds on: #24544