< Back to IRCAM Forum

Important Issues in Modalys 3.4.5.b4

Modalys beta
#1

Hi Robert !

Thanks for the new version! For me it has progressed on some aspects and - curiously enough - regressed on others.

Very cool that expression controller now is multi-dimensional. Utterly useful (I hope to prove that in a later post!)

On the downside it seems that both dynamic controller and expression controller has undergone some changes not mentioned in the readme.

In version 3.4.5.b3 dynamic controller could have its value set by default. Not possible in version 3.4.5.b4.
It is very important to fix this!

In version 3.4.5.b4 expression controller has a new “update” parameter. I assume it must the controllers update-period.
However it is not possible to give it an init value, nor are any other ways of setting it reflected in the script that remains -1.

In version 3.4.5.b3 both dynamic and expression controller had a target-attribute patrameter. Both are gone in version 3.4.5.b4.
I think this changed ought to be explained and announced. Personally I will not miss it.

The dimensionality of a dynamic controller need not always be 1. Imagine a dynamic controller setting all modal frequencies in one blow.
Such a controller must have its dimension initialised with number equal to number of modes. This is possible in LISP. But should be in Max too?

I have included to pdfs showing the above mentioned problems. One for version 3.4.5.b3 and one for version 3.4.5.b4 for easy comparison.

I hope these issues will be fixed before publication to the forum.

All the best Hans Peter

DynamicExpressionCtrl-Modalys-v345b3.pdf (1.27 MB)

#2

In addition to previous post:

Due to the weird behaviour of init-values for dynamic controller and the update-period of the expression controller (described in the previous post) even the simplest examples of expression controller linked with a dynamic controller breaks down in version 3.4.5.b4. Please see adjoint patch that compares a simple timer made using the expression controller in both version 3.4.5.b3 and b4.

It is of paramount importance that expression controller can run at sample-level for many of the nonlinear functions to work.
Of similar importance is it to be able to set init-values for dynamic controllers because many nonlinear functions depend heavily on initial states.
Progression on producing interesting nonlinear (or others) expression controller examples are not meaningful until this is fixed.

All the best Hans Peter

07F-Expression-Controller-TIMER.maxpat (81.4 KB)

#3

Hi there!

Another additional information:

I have verified the previous 07F-Expression-Controller-TIMER.maxpat in both max 6.1.9 and Max 7.3.4 in both 32 and 64 bit modes with similar results.
It thus seems that the errors are not related to the version of max. But rather are to be found in the modalys implementation?

Any help / correction is much appreciated!

Best Hans Peter

#4

… and I should add that the precious feedback instruments based on access-speed inside the expression controller no longer work in version 3.4.5.b4.
they worked in version 3.4.5.b2 but has ceased to function. I am currently stocked in all aspects of verifying and creating examples using the otherwise fantastic expression controller.

Progression is much needed.

Best Hans Peter

#5

Hi Hans Peter,

I am definitely looking into all these at the moment.

Thanks,
Robert

#6

Hi Hans Peter,

As of this evening:

  1. is fixed
  2. fixed (-1=auto, 0=every sample, otherwise a value in second ; default = 0.01 s)
  3. yes, that has been removed. It was useless.
  4. will take care of multidimensional mlys.dynamic in a future release (we need to talk about it…)

Your timer example works. Also get_value was broken, which didn’t help…

NB: the next beta is named “3.5.0.b5” (a slight version step…)

… and I should add that the precious feedback instruments based on access-speed inside the expression controller no longer work in version 3.4.5.b4. they worked in version 3.4.5.b2 but has ceased to function

Could you please an example post the corresponding patch? I have so many that I am getting lost!

Robert

#7

Yes, of course. This one for example used to be complete and fine. Now it explodes modalys halfway through the line ramps.

Best hans Peter

feedback_loop-v2.maxpat (27.5 KB)

#8

Robert!

Thanks for the new version 3.5.0.b5. I see many things are solved concerning the timer. That is great.

The feedback-loop however still does not work in version b5. It behaves as described in the previous post.
The feedback patch is what I am really into. And I sincerely hope it can be fixed very soon (before the FORUM).

Best Hans Peter

#9

Hi Hans Peter,

In order to keep it non-explosive, you need to slightly adjust the timing ramp for this example.

Robert

feedback_loop-v2_3.5.b4.maxpat (28.1 KB)

#10

Just if somebody follows this thread:

Please set update-period of the mlys.expression to zero for the example to work.
If default value of 0.01 is used it will explode. Very important!

Thanks to Robert for pointing this out.

Best Hans Peter