Struggling to understand Dorico's condensing decisions


I’d like to become more efficient in engraving, by understanding better how/why Dorico makes the condensing changes it makes. The attached file shows some of the key things that puzzle me, I was hoping someone could comment.

  1. I have a phrase with 3 horns in unisons and octaves (in a custom condensing group). Dorico decides to not condense the third horn after it, giving it a separate staff instead. Given the straightforward music, I find that puzzling, as horn 2&3 are in unison throughout, and the first horn is only in octaves. The condensing reset at the end is to ensure the final note doesn’t have any impact on this decision. Also note that a condensing reset before the phrase wouldn’t make any difference. But what is even more puzzling is that if I remove the minim in horn 1 in bar 1 (!), the condensing result from bar 2 is different. So how do notes on a different system, and separated by a phrase-break (rests) influence the condensing decision after it? Also, if a condensing reset doesn’t change this, then what exactly is a reset? I would have thought it was a ‘phrase-break’, i.e. equivalent to a rest. Is that a correct interpretation?

  2. I find the trombones also puzzling. On what basis does Dorico put different instruments in different voices (rather than single stem), if there is rhythmic unison, and the settings are “open” (single-stem unisons allowed, including mid-phrase ones). I like single-voice condensing a lot, but I often struggle to get Dorico to do it for me.

  3. NB: there’s also a labelling ‘error’ (?) at the end for the trombones, as the “3.” seems unrepresented. This is the lesser of my issues :slight_smile:

Screendump and file attached

DeleteMe.dorico (696.0 KB)

These are my “open” condensing options:

Allow single-stem unison
Allow mid-phrase unisons
Allow unlimited pitch-crossings
(Do not propagate accidentals to condensed staves)
(Require identical intervals)
Allow Amalgamation
Allow amalgamation of slurs
Allow amalgamation of playing techniques
Allow amalgamation of horizontal lines
Hide rests and label active player (at any position)
Hide rests at any position (min 2 quarter notes)
PLayers inactive for whole system: include in staff label

Lay-out options

Enable condensing
Exclude section players
NO groups excluded from condensing
Custom condensing groups:

- French Horns 1,2,3
- Trombones 1,2,3



This isn’t really an answer for you, but I personally do manual condensing overrides a lot because I’m not always able to figure out how to get the automated condensing to do what I want either. I would also like to have a better understanding of why the algorithm makes certain decisions.

We’re looking into what’s going on with the horns: it’s not doing what I would expect, in so much as allowing Horn 1 and 2 to share a stem in the first “phrase” (the initial note in horns 1 and 2) by changing their note durations to be the same is having an unexpected impact on the second phrase where the triplets begin.

In general Dorico won’t get these kinds of phrases down to a single voice because it would produce double noteheads at the unison, which it tries to avoid. The most condensed result Dorico should achieve in these situations is what you see in the trombones, where two instruments share one voice and the third ends up in its own voice. There should be stems up and down throughout the horn phrase.

On the basis of the example posted here, you don’t need to create custom condensing groups for these kinds of ensembles: Dorico will naturally put all horns and all trombones into the same condensing group in any case. I assume you must have had your reasons to do this based on some other considerations that this cut-down example doesn’t illustrate.

We’ll come back to you about the condensing in the horns once we’ve had a chance to look into it in more detail.

Thanks Daniel,

Interesting and good to know what you say about the double-note-heads being avoided. I totally understand the aesthetic rationale behind that (it does look rather clunky), but it is a somewhat counter-intuitive if the options are set to mid-phrase unisons, etc. Maybe something to turn into an option for Dorico 6.4 :wink:

I think I made the custom group for the horns to try to avoid multiple staves, and when it didn’t help I didn’t undo it. I do intuitively feel as if Dorico wants to split horns on two staves more often than with other instruments (as it’s more customary), but that’s probably my imagination.

Although you say the trombone condensing is as expected, surely it’s strange that Dorico pairs trombone 1 and 2, rather than 2 and 3, the latter two being identical? Apart from that, aLthough it’s probably not an option in Dorico right now, I would prefer to have that as single-voice, without double note-heads, but with a few labels here and there to clarify what the upper and lower notes refer to (wherever that changes). Not in the least because it tends to take less vertical space (which can be at a real premium sometimes).

In general, and also refering to mducharme’s comment, I think you don’t specifically need to fix this particular passage for my sake (I’ve already done that manually), but I would certainly welcome some more comments about how Dorico thinks (for example what you just explained). One thing, also, is a definition of what a condensing-break exactly is (or purports to be). I couldn’t a good definition in the documentation, though it’s possible I looked in the wrong place. Does a reset just undo any previous condensing change, or is it (also) always a phrase break (equivalent to a rest)? In other words, should in theory condensing at either end of a reset always be entirely uncorrelated (contrary of course to what my horn example shows)?

I might post one or two more examples of things that puzzle me, but first I need to double check it isn’t just me overlooking something.

My colleague Andrew, who is the developer behind condensing, has weighed in on this now. The issue arises from the fact that Dorico starts new phrases only at the point at which a note occurs following one or more rests. This means that rests are always considered part of the previous phrase. In this case, it means that as the second system begins, because the first and second horns cannot share a voice in the first system (because horn 1 has a half note while horn 2 has an eighth note), this limits the maximum condensing that will be allowed on the second system: horn 3 isn’t allowed on the same staff at the start of the system, because that “needs two voices”, and Dorico can’t change its mind about the maximum allowed condensing result midway through a system, so that’s that.

It can definitely be problematic that Dorico treats rests as having the same maximum allowed condensing result as the notes in the phrase, and it’s something we’d like to change in future, though it’s quite a fundamental change.

In the meantime, you can of course get the result you want (either forcing everything into a single voice, which I wouldn’t personally do because of the doubled noteheads in horns 2 and 3, or putting horns 1 and 2 into the up-stem voice and horn 3 into the down-stem voice) with a condensing change at the start of the second system.

I’m not sure exactly what you mean by “condensing-break” – that’s not a term we use.

Thanks Daniel,

I think I understand.

I apologise, I meant “condensing reset” rather than “condensing break”. For example, in the example, I tried putting a “condensing reset” at the start of the system, but it didn’t make any difference. In other words, the “rests” don’t get broken into two separate strands, which in turn causes wonder about what a condensing-reset actually is/does.

Just to drive home that point you made about Dorico avoiding double-noteheads. Am I right in concluding that a situation like this:
with a long sequence of double-voice unisons, can NOT be avoided with better options, but can only be sorted with a condensing change (such as a “reset” at the start of the solo)?

Right, when you create a condensing change, if all you do is activate the checkbox for a particular condensing group, what you’re doing is telling Dorico “a new phrase starts here”. Although it’s not the only thing Dorico considers (as mentioned above, if you are already midway through a system, if the staff setup has already been determined, simply starting a new phrase can’t change that), the start of a new phrase allows Dorico to reconsider its condensing decisions.

You don’t need to explicitly switch on the “manual condensing” slide switch and activate Reset in your condensing change: simply activating the checkbox for the condensing group you want Dorico to reconsider is sufficient.

The up-stem voice in the picture you show is a continuous phrase, because phrases only start automatically at rests; as such, if you want Dorico to be able to treat a solo passage differently but it is not preceded by a rest, you’d need a condensing change at that point.


I guess I’m still a bit puzzled why, if I tell Dorico “start a new phrase here”, even mid-way the first system (in the middle of the rests), it still makes that wrong decision on the next system. I think what you’re saying is that Dorico allocates the voices BEFORE it decides to start the new phrase (?). Anyway, it is what it is, I suppose.

Gosh, that IS an eye-opener, that you only need to tick that box. Saves a lot of scrolling, sliding and ticking!

Finally, do you have an opinion about that trombone voicing? As mentioned, I find the pairing of Tn 1+2 counter-intuitive.

Trombone 2 is neither exactly the same as trombone 1 or trombone 3, so even though it has more notes in common with trombone 3 than trombone 1, Dorico doesn’t currently consider the overall “weight” of the unisons to determine whether trombone 3 or trombone 1 would make a better partner. That would be an interesting thing for us to think about in future for phrases like this.

But… the trombone 2 and 3 ARE exactly the same…

Heh, so they are. I’ll need to double-check with Andrew tomorrow, since there is so much sophistication and complexity in the algorithms used for condensing that I may be misremembering, but as I said earlier, I don’t think Dorico considers every possible combination if it finds that 1+2 works and it can also get 3 on the staff, it doesn’t keep processing to see whether 2+3 sharing a stem would be better than 1+2 sharing a stem.

1 Like

Not sure whether I did this manually or not (probably did), but Dorico can certainly condense different passages in different ways.