Hi,
All four effects group are indeed running all the time.
I got that, but I didn't get how spatialisation of fxs comes into the play.
Are fxs producing 8 channel signals ?
Is spatialisation intertwined with fxs or done by extra processing ?
The effects are not really spatialized, strictly speaking. Essentially, there are two possible routing applied to different groups. Either the effect is mono and the out bus argument is being modulated by a routine, or there are 8 effects synths in a group, each routing to a specific loudspeaker. In the latter case they are collected in an array, over which I iterate to change the parameters of each mono effect in sequence (with the wait time somewhat randomized).
I did not know PbindFx, I will try it today.
From what I read, however, it seems to me that fx synths are created and destroyed on a per event basis? Is there a way that this process can be extended over multiple subsequent events? e.g. two events which combine into the same long-decaying short delay, which decay is then altered for both.
One fx synth (or graph of synths) is indeed applied only per event - but it would be easy to achieve what you described with Pstutter(2, ...) for the concerned fx parameter keys in PbindFx.
Removing JPverb does reduce the CPU, but changes the result quite a lot, also.
Reverb is often a crucial point in such setups. Reducing from a multichannel reverb to a multiplied/spatialised mono reverb could save a lot (if this is an acceptable option for you). From the PbindFx help these examples could give possible hints:
If I understand, this is maybe what I am doing, as explained above? I will try to incorporate PbindFx in the approach, however, as you suggest.
I will also try different kinds of reverbs and see how the CPU adapts.
Thank you for all the responses!
Regards,
V
Ex. 2d: Saving resources
Here the PbindFx concept is mixed with continously running reverb(s). This could be similar to your setup and give an idea how to combine things.
Ex. 6b: Fx synths reading audio modulation signals from external buses
This might apply to fx parameter modulation as you mentioned it.
I was experimenting with GVerb at the beginning, which seems is less intensive computationally, but it blew up on me quite a few times, when changing parameters quickly. Is there a specific reverb UGen which fares well both computationally and feedback-wise with quick changes in parameters?
I cannot say, indeed quick changes of some parameters are causing unwanted effects. But you can run several reverbs in parallel with different params and alternate routing, Ex. 2d makes such kind of compromise with two reverb units - compromise in the sense that it doesn't, wastefully, apply one reverb per event, but lets run more than only one reverb.
Regards
Daniel
-----------------------------