[Re: glissando between non-adjacent notes]

[V!ictor address@hidden]

whoops. forgot to attach the pdf.!

Thanks Han-Wen and Trevor for your responses.

On Tue, Aug 22, 2006 at 10:25:05PM -0500, Trevor Ba?a wrote:
> On 8/21/06, Han-Wen Nienhuys <address@hidden> wrote:
> >V!ctor address@hidden wrote:
> >> Currently, Lily assumes a glissando will always connect adjacent notes.
> >> (The sintax works like ties in Lily, where one puts the tie command ~
> >> after the first note)
> >> But many times one needs the glissando to span a duration that cannot be
> >> expressed with a single note, e.g. 1 + 1/4. I've attached a PDF with 2
> >> examples of this (taken from one of my own scores), together with a
> >> proposed syntax for the implementation.
> >> Essentially, these glissandi would work exactly like slurs in Lily,
> >> using one command to start a gliss and another to end it. e.g.
> >> c4\glissStart c16 f8\glissEnd
> >
> >I have the feeling that this should already more-or-less be possible
> >with current support for clusters: use a cluster for the line, and
> >blanked note heads to get the stems. Or put the blanked notes in a
> >different voice.

Haven't tried with clusters, but I have been able to get the glissandi working 
nicely using 2 voices, one for the lines, another for the durations. I've 
automated the generation of my lilypond files with python, which makes the 
production of the glissandi quite painless. 
My only complain 'till now is that its a lot more stuff going on --essentially 
twice the amount of lilypond code + a lot of warnings about clashing notes-- 
when it could be very simple (in principle). 

> >
> >Doing real timed glissandos does raise some questions: should the gliss
> >always be a straight line, or should it just follow a melodic contour?
> >If the latter, what happens with 'trilled' glissandi.

The way I see it they don't have to be straight lines. In fact I would love to 
also have splines for glissandi!! (see box 3 in my attached PDF). But glissandi 
shouldn't just follow a melodic contour. 
Glissandi are tricky. In principle one should not need to intersect a glissando 
line with durations --notes with invisible heads--. But there are two problems 
with just connecting lines between all adjacent notes:
1. Pitch quantization.
2. The staff is not a linear pitch space, but a warped space designed to 
linearly distribute a diatonic (asymmetric) scale. 

In the attached PDF, box 1, you see the following nubmers:
1. what I'd like to have in lilypond.
2. what the desired continuous line would look like if it where to be visually 
(linearly) quantized to the nearest space/line in the staff. No noteheads.
3. same as 2 but with noteheads. 
4. The closest diatonic pitch quantization (to a quarter tone) of the 
glissando. i.e. warping into the diatonic scale. 

Clearly, 2 and 3 are not useful since they misrepresent the idea of a glissando 
with constant velocity. No. 4 could be a reasonable approximation, but still 
not good enough because:
1. It's still not accurate enough. What to do in example 2 of the PDF?
2. It makes a simple gesture look more complex. 


> >
> >If the former, then this will have nasty interactions with beams: beam
> >directions are determined by stem directions, but in the case of a
> >glissando stem, the "pitch" of the stem is determined after spacing and
> >line breaking, which happens after determining directions. In this case,
> >it's still possible to do, but we would have to rewire the logic for
> >stem and beam direction code for this special case. This is tricky, so
> >it would take a couple of 100 euro.

I'm not sure I understand your explanation, but I see that the implementation 
seems to be more complex than what I thought... hmm. 

> 
> Victor & Han-Wen,
> 
> If we can figure out a good way to implement these "durated" glissandi
> (or "noncontiguous" glissandi) then I'm willing to help sponsor.
> 
> Where Han-Wen writes ...
> 
> >Another option would be to forfeit beams on glissando stems, and/or
> >require them to have preset directions.
> 
> ... I think that having stems invovled in a durated glissando having a
> preset direction would be fine ... at least it seems.
> 
> Victor - is there ever a case where the stems in a durated glissando
> should point different directions?

Well, yes, as you can see in box 1. The last two 1/4 notes would logically 
point up. *But* in the case of beamed notes, you would not want the beam to 
cross the glissando line, so in this case you would definitely want to keep 
stem directions constant. 
Let's think more about this. As we have verified already, it is possible to 
implement the non-contiguous note glissandi in Lilypond as it stands today. 
It's just not very elegant. 

Forgetting about this problem for a while, 
it would be cool to have splines for glissandi (between contiguous notes is 
OK). 
I'm going to look at the code and see if I can implement it myself! 
How much would this cost Han-Wen anyway? 

Victor.



----- End forwarded message -----

--

glissandi.pdf
Description: Adobe PDF document