Menu
▾
▴
12 Attachments
#2141 accidental spacing - some accidentals are too far from each other
Accepted
nobody
None
evidence
Enhancement
2013-09-04
2011-12-28
Anonymous
No
Originally created by: *anonymous
Originally created by: janek.li...@gmail.com
Some combinations of accidentals are put too far apart, for example in <eses' ces''>.
All possible pairs of accidentals are attached in pngs beginning with "pairs". Combinations that definately should be closer together are marked with a red dot; combinations that may look better if they were a bit closer are marked with an orange dot.
A real engraving example is also attached (more detailed examples will follow).
Related issues:
issue 2142
issue 2143
issue 2144
issue 2145
Originally posted by: janek.li...@gmail.com
More real engraving examples.
Once again a list of related issues (this time with links - previously the tracker didn't recognize issues that didn't exist at that moment)
issue 2142
issue 2143
issue 2144
issue 2145
Originally posted by: lemzw...@googlemail.com
I agree with all your suggestions except bar 75: I somehow dislike making the naturals smaller; the shape changes too much (contrary to flats and sharps).
In bar 9 of the double flats: Why gets the upper shape smaller too? I see no reason for it.
Originally posted by: janek.li...@gmail.com
Werner,
I would force naturals in sixths (m. 75) closer than they are typeset currently by Lily only when music is tight. It's possible that they don't need to be smaller in order to be closer together.
As for the upper flat in m. 9 (double flats) - I guess you're right. I shortened it so that both accidentals would be the same length, but it's a bit weird indeed.
Originally posted by: janek.li...@gmail.com
Hi,
i attach accidental pairs compiled with 2.16 and current master (with Mike's patch), for comparison.
Red ones are really bad, orange ones are too close as well but need just a bit of adjustment.
Originally posted by: PhilEHol...@googlemail.com
I actually prefer the versions created by Mike's patch in pretty much all cases. The closed-up accidentals look much neater, IMHO. Janek - I thought you were the king of wanting tighter-spaced music?
Originally posted by: dak@gnu.org
I mostly agree with Phil. Accidentals are one case where you tend to get optical clusters instead of cleanly separate objects, and that's a tradeoff in order to be able to take them in faster. Personally, I would think it rather desirable that the arrangement of that cluster tried to match the shape of the corresponding notehead cluster, where possible.
But I don't know what the references have to say about that.
Originally posted by: janek.li...@gmail.com
Oh, i definitely want tightly spaced accidentals when the music itself is spaced tightly. When there's lots of space available, i'd prefer having them a bit farther apart.
David, you do have a point with accidentals forming clusters that may be easier to read if they are tight. But i'm not totally sure.
I've checked some engravings and they seem to be a bit inconsistent in this regard. Some placed accidentals very closely even if there was a lot of space available. Others did what i suggest - you may for example look at the image in this comment: https://code.google.com/p/lilypond/issues/detail?id=2143#c7
Originally posted by: k-ohara5...@oco.net
Ross says "Mulitple, unlike accidentals may be drawn closer together [than the accidentals in key signatures] but preferably without touching"
Originally posted by: lemzw...@googlemail.com
Looking at Janek's PDF files, all accidental positions except in bars 7, 25, 75, 85, 105, and 134 (which are *a bit* too close) are fine for me. A special case is bar 145; IMHO the correct position is exactly in the middle of what is shown in the 2.16 PDF file and the PDF with Mike's patch.
Originally posted by: janek.li...@gmail.com
You really don't think that flat and natural in fourth (m.13 and 43) aren't too close? I think this is actually the worst problem, and i'm afraid they will become illegible in small font size. I don't think i've seen an engraving with these accidentals so close - in Ross' example posted above they are noticeably farther away.
Originally posted by: PhilEHol...@googlemail.com
In the example Keith posted, which I assume are from Ross, it seems to me that the flat-nat in 4ths are touching, which in the patched version they don't. TBH, I like all the ones in the new version.
Originally posted by: milimet...@gmail.com
(No comment was entered for this change.)
Cc: -milimet...@gmail.com
Originally posted by: lemzw...@googlemail.com
@janek: Bars 13 and 43 are OK for me, since the large vertical size of the flat makes them clearly distinct.
Originally posted by: tdanielsmusic
I'm happy with the bars highlighted in Mike's patch, but why the
difference in the top accidental in the bar pairs 11-12, 17-18,
23-24? The first bar in each of these pairs looks wrong to me.
Trevor
Originally posted by: k-ohara5...@oco.net
Note-heads separated as far as octaves have their accidentals left-aligned, both before and after Mike's patch. The ones that look wrong to Trevor are where the accidentals are different, which is the unusual case, making the octave into some dissonant interval.
I don't know why Janek hasn't pursued the horizon_padding (the kind of padding that prevents interleaving) that I suggested in the code review. For me, it resolves the tight spacing in measures 13 19 43 49 61 67 75 81 85 105 111 134 and 140. It also moves the double-sharp right of the double-flat in measure 27, but I don't think that will trouble anyone.
Originally posted by: janek.li...@gmail.com
@ Keith (comment #15):
i'm sorry that i hadn't checked the results with horizon_padding earlier. I did so now and the results, while not perfect, are definitely acceptable. As you said, it solves too tight spacing in many places (see attachment). In my opnion, this is the best accidental spacing of all that i've seen, and we can make it default.
@ Phil (https://code.google.com/p/lilypond/issues/detail?id=2141#c11):
I believe that you are using wrong metric. In other words, the fact that flat and natural in Ross example touch doesn't necessarily mean that they are visually closer than accidentals spaced with Mike's patch (which don't touch). For example, you could measure accidentals' "distance" by area between them, or by the difference between their left edges. If we choose one of these methods (and they give more "natural" results in my opinion), we'll see that accidentals from Ross are farther apart - see attached.
Is it clear what i mean here?
Originally posted by: k-ohara5...@oco.net
I noticed that a few of the accidental pairs mentioned above (naturals in a sixth for example) were hard to read in print at normal distance.
Just a tiny bit of padding separates those few pairs of accidentals that are hard to read (see attached).
http://codereview.appspot.com/13223043
Labels: Patch-new
Cc: -mts...@gmail.com
Originally posted by: pkx1...@gmail.com
Passes Make, make check and full make doc.
Reg test diffs here:
https://www.hightail.com/download/bWJwMFhpVnNEa1ZBSXRVag
Labels: -Patch-new Patch-review
Owner: k-ohara5...@oco.net
Cc: mts...@gmail.com
Originally posted by: janek.li...@gmail.com
Keith,
very good!
There are some places where the spacing isn't perfect yet (in my opinion), but the change is definitely and improvement. I like how flat and natural in a fourth (m.43) aren't cramped now.
Thanks a lot
Originally posted by: pkx1...@gmail.com
Patch on countdown for September 3rd - 06:00 GMT
Labels: -Patch-review Patch-countdown
Originally posted by: janek.li...@gmail.com
After pushing the patch, do you intend to close the issue?
As i think that in some cases the spacing could be improved even further, it may make sense to leave it open.
Originally posted by: pkx1...@gmail.com
Patch counted down - please push
Labels: -Patch-countdown Patch-push
Originally posted by: k-ohara5...@oco.net
Commit [r202d2afd0a4c6f199e063148d6e7368868db9390] produces output (comment 17) that puts accidentals as close as in the requested behavior (comment 1) except for measures 3 39 141 and 145 (and combinations that do not arise in music in measures 33 116 and 27).
Labels: -Patch-push Needs-evidence
Owner: ---