Just brainstorming here: maybe getting the absolute value of the
change in panning position, and sending that through portk, and adding
it to your dispersion? Something like:

kprev init 0
kdelta = abs(kposition - kprev)
kprev = kposition
kexpand portk kdelta, khtime
kspread = kbasespread+(kexpand * kresiduewidth)

this is a little simplistic, but may be enough?

On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
<peimankhosravi@gmail.com> wrote:
> Hello,
>
> I'm working on making a multichannel granular texture instrument with
> partikkel (or even just an impulse generator like gausstrig) with a twist:
> each time the panning 'position' is changed the sound leaves a small traces
> behind, with a user-defined decay rate (e.g. 10 seconds with an exponential
> decay). Any idea how this could be realised in an efficient way? I suppose
> the design would need to be similar to the attack/decay duration of a
> compressor but I've no idea how that's done.
>
> Many Thanks
> Peiman

Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Hi Justin,

Thanks for the idea. Yes something like that would work, but I was hoping for a little bit more control over the morphology of the dispersion. Exponential seems works better. I came up with this test csd. Not there yet but it wort of works in principle I think.

Best,
Peiman

<CsoundSynthesizer>

<CsOptions>
-W -odevaudio -d
</CsOptions>

<CsInstruments>
sr     = 44100
ksmps  = 1
nchnls = 2




    instr 1 ;Simple sine at 440Hz

aout    oscil 1000, 2000, 1

kpan    oscil .5, .5, 1
kpan = kpan +.5

if (kpan==1) then
    ktest=0
    reinit decay
elseif (kpan==0) then
    ktest=1
endif

if (ktest==1) then
kexpon = kpan
goto play
endif


decay:
kexpon    expon    1, 3, 0.000001
rireturn


play:
kexpon port kexpon, .01
aexpon upsamp kexpon

    outs    aout*aexpon, aout*aexpon        


;printk2 ktest

    endin

</CsInstruments>

<CsScore>
f1 0 4096 10 1
i1 0 10
</CsScore>

</CsoundSynthesizer>

On 4 August 2012 13:58, Justin Smith <noisesmith@gmail.com> wrote:

Just brainstorming here: maybe getting the absolute value of the
change in panning position, and sending that through portk, and adding
it to your dispersion? Something like:

kprev init 0
kdelta = abs(kposition - kprev)
kprev = kposition
kexpand portk kdelta, khtime
kspread = kbasespread+(kexpand * kresiduewidth)

this is a little simplistic, but may be enough?

On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
<peimankhosravi@gmail.com> wrote:
> Hello,
>
> I'm working on making a multichannel granular texture instrument with
> partikkel (or even just an impulse generator like gausstrig) with a twist:
> each time the panning 'position' is changed the sound leaves a small traces
> behind, with a user-defined decay rate (e.g. 10 seconds with an exponential
> decay). Any idea how this could be realised in an efficient way? I suppose
> the design would need to be similar to the attack/decay duration of a
> compressor but I've no idea how that's done.
>
> Many Thanks
> Peiman

Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Just brainstorming here: maybe getting the absolute value of the
change in panning position, and sending that through portk, and adding
it to your dispersion? Something like:

kprev init 0
kdelta = abs(kposition - kprev)
kprev = kposition
kexpand portk kdelta, khtime
kspread = kbasespread+(kexpand * kresiduewidth)

this is a little simplistic, but may be enough?

On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
<peimankhosravi@gmail.com> wrote:
> Hello,
>
> I'm working on making a multichannel granular texture instrument with
> partikkel (or even just an impulse generator like gausstrig) with a twist:
> each time the panning 'position' is changed the sound leaves a small traces
> behind, with a user-defined decay rate (e.g. 10 seconds with an exponential
> decay). Any idea how this could be realised in an efficient way? I suppose
> the design would need to be similar to the attack/decay duration of a
> compressor but I've no idea how that's done.
>
> Many Thanks
> Peiman

Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

OK I'm taking another look at your idea. What I was doing was way too convoluted.

Can you walk me through this formula please?

Is kbasespread for vbap?

Thanks again
Peiman

On 4 August 2012 13:58, Justin Smith <noisesmith@gmail.com> wrote:

Just brainstorming here: maybe getting the absolute value of the
change in panning position, and sending that through portk, and adding
it to your dispersion? Something like:

kprev init 0
kdelta = abs(kposition - kprev)
kprev = kposition
kexpand portk kdelta, khtime
kspread = kbasespread+(kexpand * kresiduewidth)

this is a little simplistic, but may be enough?

On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
<peimankhosravi@gmail.com> wrote:
> Hello,
>
> I'm working on making a multichannel granular texture instrument with
> partikkel (or even just an impulse generator like gausstrig) with a twist:
> each time the panning 'position' is changed the sound leaves a small traces
> behind, with a user-defined decay rate (e.g. 10 seconds with an exponential
> decay). Any idea how this could be realised in an efficient way? I suppose
> the design would need to be similar to the attack/decay duration of a
> compressor but I've no idea how that's done.
>
> Many Thanks
> Peiman

Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Yeah, I had imagined you could just make the spread be a slowly
updating function of the movement, and that might create an impression
of a trace left by a moving sound.

If you want the output traces to follow old output, you could also put
a reverb in (at the instrument level), on each of the vbap outputs
before writing the outptut. Combining with my previous suggestion we
get something like:

kprev_az init 0
kdelta = abs (kazim-kprev_az)
kprev_az = kazim
kd portk kdelta, 3
kspread  kd*ispreadfactor ; instead of this, you could do an
exponential or table lookup based on kdelta

ar1, ar2, ar3, ar4 vbap4 asig, kazim, kelev, kspread

at1 nreverb ar1, 10, .8
at2 nreverb ar2, 10, .8
at3 nreverb ar3, 10, .8
at4 nreverb ar4, 10, .8

ktlv = ampdb(kd*itracefactor - 30)

outq ar1+at1*ktlv, ar2+at2*ktlv, ar3+at3*ktlv, ar4+at4*ktlv


It occurs to me that for this kind of effect, you could save a bunch
of cpu by getting not audio signals as output of vbap, but amplitude
scales for each channel, then the reverb here would only need to be
calculated once

Otherwise you could use schedkwhen to periodically launch a granulator
with a snapshot of the current parameters while pan is changing. This
would be a sort of meta-granulation if you put a volume envelope on
each granulator.

On Sat, Aug 4, 2012 at 1:02 PM, peiman khosravi
<peimankhosravi@gmail.com> wrote:
> Another issue is that the instrument parameters (e.g. periodicity or grain
> density) will be changing as the sound 'moves' but the trace would need to
> remain unchanged. So I was thinking I need to have two copies of the
> instrument running in parallel and sample the variable at a given moment in
> time. Maybe I'll need to have a break-point function or something to define
> the times at which the variables are sampled. The dispersion could also have
> an effect on the variables (so for instance dispersion by decreasing grain
> density may be more effective than simple amplitude decay). Need to think
> about it a bit more. Annoyingly this is how I do it in protools but it would
> be much more efficient to programme it in instead of spending hours tweaking
> automation!
>
> P
>
> On 4 August 2012 20:55, peiman khosravi <peimankhosravi@gmail.com> wrote:
>>
>> OK I'm taking another look at your idea. What I was doing was way too
>> convoluted.
>>
>> Can you walk me through this formula please?
>>
>> Is kbasespread for vbap?
>>
>> Thanks again
>> Peiman
>>
>> On 4 August 2012 13:58, Justin Smith <noisesmith@gmail.com> wrote:
>>>
>>> Just brainstorming here: maybe getting the absolute value of the
>>> change in panning position, and sending that through portk, and adding
>>> it to your dispersion? Something like:
>>>
>>> kprev init 0
>>> kdelta = abs(kposition - kprev)
>>> kprev = kposition
>>> kexpand portk kdelta, khtime
>>> kspread = kbasespread+(kexpand * kresiduewidth)
>>>
>>> this is a little simplistic, but may be enough?
>>>
>>> On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
>>> <peimankhosravi@gmail.com> wrote:
>>> > Hello,
>>> >
>>> > I'm working on making a multichannel granular texture instrument with
>>> > partikkel (or even just an impulse generator like gausstrig) with a
>>> > twist:
>>> > each time the panning 'position' is changed the sound leaves a small
>>> > traces
>>> > behind, with a user-defined decay rate (e.g. 10 seconds with an
>>> > exponential
>>> > decay). Any idea how this could be realised in an efficient way? I
>>> > suppose
>>> > the design would need to be similar to the attack/decay duration of a
>>> > compressor but I've no idea how that's done.
>>> >
>>> > Many Thanks
>>> > Peiman
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>             https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>> csound"
>>>
>>
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

It just seems like separation of concerns to me - vbap is already
there for spreading signals between speakers, why use another method
as well?

And when I put it that way, perhaps instead of changing the spread you
could use a second instance of vbap, where its azimuth input is run
through portk and maybe even a delay line (still using the reverb also
if you want lingering artifacts). Or maybe just the reverb without a
second vbap or a change in spread would be best.

On Sat, Aug 4, 2012 at 1:56 PM, peiman khosravi

<peimankhosravi@gmail.com> wrote:
> OK I see, this code is really useful, thanks.
>
> Could you just clarify why are you using the spread factor here? Wouldn't it
> be sufficient to deal with each speaker independently?
>
> Thanks
> Peiman
>
>
>
> On 4 August 2012 21:34, Justin Smith <noisesmith@gmail.com> wrote:
>>
>> Yeah, I had imagined you could just make the spread be a slowly
>> updating function of the movement, and that might create an impression
>> of a trace left by a moving sound.
>>
>> If you want the output traces to follow old output, you could also put
>> a reverb in (at the instrument level), on each of the vbap outputs
>> before writing the outptut. Combining with my previous suggestion we
>> get something like:
>>
>> kprev_az init 0
>> kdelta = abs (kazim-kprev_az)
>> kprev_az = kazim
>> kd portk kdelta, 3
>> kspread  kd*ispreadfactor ; instead of this, you could do an
>> exponential or table lookup based on kdelta
>>
>> ar1, ar2, ar3, ar4 vbap4 asig, kazim, kelev, kspread
>>
>> at1 nreverb ar1, 10, .8
>> at2 nreverb ar2, 10, .8
>> at3 nreverb ar3, 10, .8
>> at4 nreverb ar4, 10, .8
>>
>> ktlv = ampdb(kd*itracefactor - 30)
>>
>> outq ar1+at1*ktlv, ar2+at2*ktlv, ar3+at3*ktlv, ar4+at4*ktlv
>>
>>
>> It occurs to me that for this kind of effect, you could save a bunch
>> of cpu by getting not audio signals as output of vbap, but amplitude
>> scales for each channel, then the reverb here would only need to be
>> calculated once
>>
>> Otherwise you could use schedkwhen to periodically launch a granulator
>> with a snapshot of the current parameters while pan is changing. This
>> would be a sort of meta-granulation if you put a volume envelope on
>> each granulator.
>>
>>
>> On Sat, Aug 4, 2012 at 1:02 PM, peiman khosravi
>> <peimankhosravi@gmail.com> wrote:
>> > Another issue is that the instrument parameters (e.g. periodicity or
>> > grain
>> > density) will be changing as the sound 'moves' but the trace would need
>> > to
>> > remain unchanged. So I was thinking I need to have two copies of the
>> > instrument running in parallel and sample the variable at a given moment
>> > in
>> > time. Maybe I'll need to have a break-point function or something to
>> > define
>> > the times at which the variables are sampled. The dispersion could also
>> > have
>> > an effect on the variables (so for instance dispersion by decreasing
>> > grain
>> > density may be more effective than simple amplitude decay). Need to
>> > think
>> > about it a bit more. Annoyingly this is how I do it in protools but it
>> > would
>> > be much more efficient to programme it in instead of spending hours
>> > tweaking
>> > automation!
>> >
>> > P
>> >
>> > On 4 August 2012 20:55, peiman khosravi <peimankhosravi@gmail.com>
>> > wrote:
>> >>
>> >> OK I'm taking another look at your idea. What I was doing was way too
>> >> convoluted.
>> >>
>> >> Can you walk me through this formula please?
>> >>
>> >> Is kbasespread for vbap?
>> >>
>> >> Thanks again
>> >> Peiman
>> >>
>> >> On 4 August 2012 13:58, Justin Smith <noisesmith@gmail.com> wrote:
>> >>>
>> >>> Just brainstorming here: maybe getting the absolute value of the
>> >>> change in panning position, and sending that through portk, and adding
>> >>> it to your dispersion? Something like:
>> >>>
>> >>> kprev init 0
>> >>> kdelta = abs(kposition - kprev)
>> >>> kprev = kposition
>> >>> kexpand portk kdelta, khtime
>> >>> kspread = kbasespread+(kexpand * kresiduewidth)
>> >>>
>> >>> this is a little simplistic, but may be enough?
>> >>>
>> >>> On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
>> >>> <peimankhosravi@gmail.com> wrote:
>> >>> > Hello,
>> >>> >
>> >>> > I'm working on making a multichannel granular texture instrument
>> >>> > with
>> >>> > partikkel (or even just an impulse generator like gausstrig) with a
>> >>> > twist:
>> >>> > each time the panning 'position' is changed the sound leaves a small
>> >>> > traces
>> >>> > behind, with a user-defined decay rate (e.g. 10 seconds with an
>> >>> > exponential
>> >>> > decay). Any idea how this could be realised in an efficient way? I
>> >>> > suppose
>> >>> > the design would need to be similar to the attack/decay duration of
>> >>> > a
>> >>> > compressor but I've no idea how that's done.
>> >>> >
>> >>> > Many Thanks
>> >>> > Peiman
>> >>>
>> >>>
>> >>> Send bugs reports to the Sourceforge bug tracker
>> >>>
>> >>> https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> >>> Discussions of bugs and features can be posted here
>> >>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>> >>> "unsubscribe
>> >>> csound"
>> >>>
>> >>
>> >
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>             https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

But if the sound is just leaving behind a trace then the trace would only be in one speaker and not more spread. I feel like I'm missing something though (?).

P

On 4 August 2012 22:09, Justin Smith <noisesmith@gmail.com> wrote:

It just seems like separation of concerns to me - vbap is already
there for spreading signals between speakers, why use another method
as well?

And when I put it that way, perhaps instead of changing the spread you
could use a second instance of vbap, where its azimuth input is run
through portk and maybe even a delay line (still using the reverb also
if you want lingering artifacts). Or maybe just the reverb without a
second vbap or a change in spread would be best.

On Sat, Aug 4, 2012 at 1:56 PM, peiman khosravi

<peimankhosravi@gmail.com> wrote:
> OK I see, this code is really useful, thanks.
>
> Could you just clarify why are you using the spread factor here? Wouldn't it
> be sufficient to deal with each speaker independently?
>
> Thanks
> Peiman
>
>
>
> On 4 August 2012 21:34, Justin Smith <noisesmith@gmail.com> wrote:
>>
>> Yeah, I had imagined you could just make the spread be a slowly
>> updating function of the movement, and that might create an impression
>> of a trace left by a moving sound.
>>
>> If you want the output traces to follow old output, you could also put
>> a reverb in (at the instrument level), on each of the vbap outputs
>> before writing the outptut. Combining with my previous suggestion we
>> get something like:
>>
>> kprev_az init 0
>> kdelta = abs (kazim-kprev_az)
>> kprev_az = kazim
>> kd portk kdelta, 3
>> kspread  kd*ispreadfactor ; instead of this, you could do an
>> exponential or table lookup based on kdelta
>>
>> ar1, ar2, ar3, ar4 vbap4 asig, kazim, kelev, kspread
>>
>> at1 nreverb ar1, 10, .8
>> at2 nreverb ar2, 10, .8
>> at3 nreverb ar3, 10, .8
>> at4 nreverb ar4, 10, .8
>>
>> ktlv = ampdb(kd*itracefactor - 30)
>>
>> outq ar1+at1*ktlv, ar2+at2*ktlv, ar3+at3*ktlv, ar4+at4*ktlv
>>
>>
>> It occurs to me that for this kind of effect, you could save a bunch
>> of cpu by getting not audio signals as output of vbap, but amplitude
>> scales for each channel, then the reverb here would only need to be
>> calculated once
>>
>> Otherwise you could use schedkwhen to periodically launch a granulator
>> with a snapshot of the current parameters while pan is changing. This
>> would be a sort of meta-granulation if you put a volume envelope on
>> each granulator.
>>
>>
>> On Sat, Aug 4, 2012 at 1:02 PM, peiman khosravi
>> <peimankhosravi@gmail.com> wrote:
>> > Another issue is that the instrument parameters (e.g. periodicity or
>> > grain
>> > density) will be changing as the sound 'moves' but the trace would need
>> > to
>> > remain unchanged. So I was thinking I need to have two copies of the
>> > instrument running in parallel and sample the variable at a given moment
>> > in
>> > time. Maybe I'll need to have a break-point function or something to
>> > define
>> > the times at which the variables are sampled. The dispersion could also
>> > have
>> > an effect on the variables (so for instance dispersion by decreasing
>> > grain
>> > density may be more effective than simple amplitude decay). Need to
>> > think
>> > about it a bit more. Annoyingly this is how I do it in protools but it
>> > would
>> > be much more efficient to programme it in instead of spending hours
>> > tweaking
>> > automation!
>> >
>> > P
>> >
>> > On 4 August 2012 20:55, peiman khosravi <peimankhosravi@gmail.com>
>> > wrote:
>> >>
>> >> OK I'm taking another look at your idea. What I was doing was way too
>> >> convoluted.
>> >>
>> >> Can you walk me through this formula please?
>> >>
>> >> Is kbasespread for vbap?
>> >>
>> >> Thanks again
>> >> Peiman
>> >>
>> >> On 4 August 2012 13:58, Justin Smith <noisesmith@gmail.com> wrote:
>> >>>
>> >>> Just brainstorming here: maybe getting the absolute value of the
>> >>> change in panning position, and sending that through portk, and adding
>> >>> it to your dispersion? Something like:
>> >>>
>> >>> kprev init 0
>> >>> kdelta = abs(kposition - kprev)
>> >>> kprev = kposition
>> >>> kexpand portk kdelta, khtime
>> >>> kspread = kbasespread+(kexpand * kresiduewidth)
>> >>>
>> >>> this is a little simplistic, but may be enough?
>> >>>
>> >>> On Sat, Aug 4, 2012 at 4:46 AM, peiman khosravi
>> >>> <peimankhosravi@gmail.com> wrote:
>> >>> > Hello,
>> >>> >
>> >>> > I'm working on making a multichannel granular texture instrument
>> >>> > with
>> >>> > partikkel (or even just an impulse generator like gausstrig) with a
>> >>> > twist:
>> >>> > each time the panning 'position' is changed the sound leaves a small
>> >>> > traces
>> >>> > behind, with a user-defined decay rate (e.g. 10 seconds with an
>> >>> > exponential
>> >>> > decay). Any idea how this could be realised in an efficient way? I
>> >>> > suppose
>> >>> > the design would need to be similar to the attack/decay duration of
>> >>> > a
>> >>> > compressor but I've no idea how that's done.
>> >>> >
>> >>> > Many Thanks
>> >>> > Peiman
>> >>>
>> >>>
>> >>> Send bugs reports to the Sourceforge bug tracker
>> >>>
>> >>> https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> >>> Discussions of bugs and features can be posted here
>> >>> To unsubscribe, send email sympa@lists.bath.ac.uk with body
>> >>> "unsubscribe
>> >>> csound"
>> >>>
>> >>
>> >
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>             https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"