Semi-random stem direction?

I don’t understand why Dorico sometimes changes stem direction in a way which doesn’t follow the ‘rules’. I can understand trying to keep stems in a passage more or less consistent but often it just looks wrong. Take a look at this cello part. Why are there down stems in bar 1? The stems in bars 18, 32 and 34 should officially go up although I can imagine down stems looking better in the context but what about bars 59 and 68? To me these really look wrong. FWIW, midline stem direction is set to be down. Dorico shouldn’t be making decisions like these!

Did you input this part from a blank project, or is it imported?

If it’s the latter, you could select all and click Reset.

Nope. All done within Dorico. To make matters worse, flipping the stem direction in the score can’t be propagated, so I have to do it in the part, as well.

Have you checked voice colors? Can’t imagine that would be an issue here, but…

I’ve never seen something that.

I re-entered the snippets you mentioned into a new score, and the stem directions seemed to be the same as yours everywhere except bar 1.

The “obvious” explanation is that you flipped the stems in bar 1 by accident while trying to do something else, of course.

Did you try Edit / Reset Position and Edit / Reset Appearance on bar 1?

…and Edit > Stem > Removed Forced Stem, just in case stem directions are something that Reset Position/Design doesn’t reset…

@ Dan; no, they’re all voice 1.
@ Rob & Leo; I’m sure I didn’t flip the stems in bar 1 by mistake, and that doesn’t explain the strange stem directions in the other bars. Neither Reset Position nor Reset Appearance did anything and Remove Forced Stem just returned everything to the way it was, i.e. in the example above.

I really hope someone from Steinberg chimes in here. This isn’t the first time I’ve noticed aberrant stem directions.

Hi Vaughan!

Engraving options>Notes>Stems>Stem Direction>Determine by context will get you the proper results as in the following jpg:

Do be aware that this setting will sometimes create measures where two notes on the middle line, separated by a rest, will have different stem directions because of context (note notes preceding the first note of the measure, and the notes following the note after the rest.

Gee, you even entered the whole cello part! (Streams of 8ths do go quickly, though, don’t they?) I know about the stem direction setting and, except in unusual circumstances, I don’t want the stem direction to be determined by context. The example in the Engraving Options window just looks wrong to me. Still, it doesn’t make sense. You might expect the stems in bb. 1 and 18 to flip down in the context but certainly not according to the default direction, whereas the opposite happens, and it still doesn’t explain the strange stem direction in bb. 59, 60, 66 and 68.

… I scanned the png and had PhotoScore read it. I’m dedicated, but there are limits!

Mmm, not bad. It only missed the tie between bb. 58 and 59. I think I still have a version of it somewhere. Perhaps I should give it another chance. I tried it years ago when I had to make parts for a lot of Bach cantatas but I discovered that it was almost as quick and definitely more secure to play everything in than to scan it and have to check everything.

Honestly, it depends on the source. I have some very good results sometimes, and I defined a Stream Deck configuration that really helps me to go faster on Photoscore Ultimate. It’s been almost a year since last update, I really hope they’re still working and that they’re going to bring us some good stuff (the find and replace function is nice but bugged, so a real debugging would be great).

Very often true. But there are jobs, especially with a clean copy, where it can really shine and save me oodles of time. By now I tend to know in advance which method I should use on a case-per-case basis.

Vaughan, fwiw, I actually think your original post looks pretty good. Mm. 59, for instance, Dorico is clearly trying to keep the same pattern as mm. 60; 60 is down because of the figure it is tied to in 61. Mm. 18 is to keep a single figure from being the opposite direction as all the others that comes after it. I’d bet money that engravers of old who did this by hand would have made the same choice. Honestly, I’d rather read a few patterns all going the same direction than have stems flip flopping all over the place but be “technically” correct. Now, I grant fully that there is some rule bending and this is my personal preference so I’m not saying that you’re wrong for wanting it the other way. But I don’t think the default looks bad at all; in fact, I think it looks quite pleasing. Gould mentions on page 24 that when there are unequal numbers of notes which weigh in either side of center, the “majority of the stems go in the right direction.” I’m guessing it is this clause that is at play here for some of the measures you pointed out. (68 for example)

Ah-- one other thing, she also says that “the exception is when a minority of notes are much further from the middle stave-line than the majority. In this case it is better to have a majority of shorter stems in the ‘wrong’ direction, rather than many long stems.”

I find it interesting that she says ‘wrong’ (quotation marks are original); this indicates there is genuinely some fluidity in the matter.

I tend to agree with Romanos. Whether or not it follows some set of “rules”, apart form the first bar it looks OK. The only thing that looked strange to me was the down stems after the long tie in bar 59, but after flipping the stems in bar 59 I think I prefer the way Dorico did it, on the basis of “does it look nice” not “is it theoretically correct”.

I agree completely about aesthetics trumping strict rules, although where one draws the line can occasionally place one on a slippery slope. I also agree that Dorico’s algorithms for determining stem direction are pretty sophisticated, also taking the length of the majority of the stems into account. Much of this discussion is a matter of opinion, too. Bar 59 looks strange but flipping it makes it inconsistent with bar 60. However flipping both makes 60/61 more consistent with 62/63. Etc.

What bothers me is bar 1, which is wrong in pretty much any universe, especially taking Dorico’s stem direction settings into account. If you have stem direction set to Determine by context, it will place those stems up, which is both aesthetically pleasing and theoretically correct, but inconsistent in the context of the bars following it. If you have stem direction set to Default, it flips them down which doesn’t make any sense at all. This is reproducible.

If you enter this from scratch with Engraving Options set to factory settings, you do NOT get stems down in bar 1 by default. Or at least, I don’t.

Factory settings are as follows:

If you change the top option to “Use default direction” and you don’t change the second option, then Dorico’s respecting your choice: There are two notes on the middle line and one underneath it, so the two notes the middle line win the argument and force the stems down. That’s surely your decision.

Concerning the second option, shouldn’t ‘examination of surrounding notes’ take the B-flat in between into account?