I should add diffusion can use one channel successfully over several
speakers for creative effect - but her eIm talking about the accuracy of
panning, point sources etc
So I tried to accurately describe how panning is done... đ
Stefan
On Sun, 25 Oct 2020 at 09:32, Augustine Leudar <augustineleu...@gmail.com>
wrote:
Stefan - I think we may be getting crossed wires here (pun not intended)
Me: *âone channel over several speakers sounds crapâ*
"
You: *"*
*I would say your statement is just wrong. We are following normal
panning rules, which are proven in millions of *
*recordings."*
Stereo recordings have two channels, not one - unless you're referring to
mono which of course can sound fine with one channel on two speakers - but
has no panning whatsoever
But perhaps you are not understanding what I am saying or perhaps you are
trolling me :) You suggest "just trying VBAP - well I first "just tried
VBAP" fifteen years ago and have used it in literally hundreds of
installations since and much prefer it to ambisonics in 90 percent of
situations.- I literally use it for work nearly every day and its cousin
DBAP and other forms of amplitude panning, day in day out so I do hope you
take the time to actually understand what I am trying to say otherwise I
fear this may be wasted time for both of us .
"*Stereo to 4 speakers: You canât map stereo to positions which are out
of the stereo front*."
You can and people do - but it sounds crap - which was my point - people
upmix a stereo file to 5.1 for example in post houses all the time there is
a variety of ways they do this and you can read about them here :
https://link.springer.com/chapter/10.1007/978-3-642-10546-3_28
"*It is important to see that every position is panned to 2 speakers in *
*2D, and (usually) 3 speakers in 3D.*"
????? Yes and no - stereo is panned to positions in 2 speakers of course
but I would think more carefully about what 2D and 3D actually mean .......
technically speaking - 3D just refers to 3 dimensions. 3D could refer to a
million speakers, so could 2D actually. It's very rare for any kind of
spatial audio to be rendered over 3 speakers though it happens. Whilst I
know simple surround systems are referred to as 3D In reality the whole
thing is a misnomer, 3D should include height and proximity. So a point
source (1 speaker ) should be called 0D, Stereo should be called 1D - a
line - quad/5.1 octaphonic/3 speakers you refer to should be called 2D as
its just a flat surface- from then on we include height - Ambisonics when
rendered with height, should, in my opinion, should be called 2.5 D as it
cant really create proximity properly - aka sounds coming close to you.
True 3D audio, where a sound can be anywhere in 3D space including sounds
that come right up close to you can only be created by DBAP in my
experience, or other amplitude panning and perhaps binaural (not heard
this convincingly yet) though Matt Montags WFS with height system might be
able to do it. More on the definition of dimensions :
https://gadgetsthink.com/what-is-dimension-full-explained-1d-2d-3d-4d-etc/
You could argue you could add height to 3 (or even 2 or 1) speakers using
psychoacoustic effects , directional bands etc making it 3D - or you could
stick one speaker up a lampost for height - but that's another can of
worms. Recent "developments" in 4D, 8D audio etc etc are just marketing
gimmicks.
Anyway, - I am not talking about ambisonics or object-based panning (and
yes object-based panning systems such as Atmos use amplitude panning in a
system similar to VBAP but instead of a triangle of speakers uses a
rectangular tesselation :
https://patents.google.com/patent/WO2016210174A1/en) .
I am talking for want of a better expression amplitude panning and also
the Scheoops type system at the beginning of this thread.
Lets me try and explain things as clearly as possible.
IF I have a stereo file in which a bird flies from one speaker to another
and I simply "upmix" this stereo file by doubling it onto a quad array then
the bird will be flying from the front left to the front right speaker AND
the back left to the back right speaker - which willl sound crap, or at
least not usually the desired effect . So in the example of the Scheops
setup - you have 8 hypercardioid speakers point to the 8 corners of the
cube - each one of those channels is meant to go go to ONE speaker of a
cube array. Ie each one of those mics and speakers covers one corner or an
eighth of the 3D sound field. Let's go back to the bird. Say I have a bird
happily staying in the top right speaker of my octaphonic cube as was
recorded there by my mic or positioned there by mu panner. If you
suddenly get that top right corner channel of the octatonic cube and put it
on two speakers instead of just the one it was originally meant to
represent (aka upmixing) then the bird will no longer be an accurate point
source in one place it will now be coming from two places, if those two
speakers are in different trees - that bird will now be coming from two
trees - aka the bird will suddenly be larger, or the imaging blurred and
you might get lucky with precedence effect,this will apply to all panning
in the sound scene too if you have upmixed all speakers, the further apart
the speakers are the worse it will be. For me this kind of thing is
crucially important as I do walkaround installations that cannot have a
"sweet spot". So upmixing presents a problem for this type of 3D audio
recordings - and perhaps less so for ambisonics. Now downmixing with this
kind of recordings and composition actually works quite well and I have
tried it many times when I've needed to send a rough mix to prospective
clients and I am left with the daunting task of downmixing a 28 channel
installation to a stereo file - you have to be very selective about which
channels you use and the perspective of the listener. Funnily enough the
stereo recordings I have downmixed from quad and octaphonic recordings
sound strangely spacious. I hope this makes my point of view clear - it is
quite simple and based on years of experience and research and is quite
practical - things have to work or I don't get gigs and the public response
is not good simple as that.
Delightful as this conversation is I have to get back to work and will be
offline for a bit - I bid you good day sir.
On Sun, 25 Oct 2020 at 01:34, Stefan Schreiber <st...@mail.telepac.pt>
wrote:
Stereo to 4 speakers: You canât map stereo to positions which are out
of the stereo front.
Octomic to 20 speakers: Should actually (and does) work, via simple
panning.
It is important to see that every position is panned to 2 speakers in
2D, and (usually) 3 speakers in 3D.
(If speakers should stay empty I donât see any problem.)
âone channel over several speakers sounds crapâ
We are following normal panning rules, which are proven in millions of
recordings. So I would say your statement is just wrong.
You donât spread â8 speakers over 20 speakersâ in some statistical
ways, if that is what you meant.
So I just have suggested to âtryâ VBAP, obtaining some very reasonable
(and proven) results.
Object and speaker panning is not very different, by the way.
(You can see the speakers also as objects. )
Best,
Stefan
----- Mensagem de Augustine Leudar <augustineleu...@gmail.com> ---------
Data: Fri, 23 Oct 2020 23:21:50 +0100
De: Augustine Leudar <augustineleu...@gmail.com>
Assunto: Re: [Sursound] Recorder for ORTF-3D OUTDOOR SET
Para: Surround Sound discussion group <sursound@music.vt.edu>
...
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit
account or options, view archives and so on.
