Voxengo Premium Membership - All Voxengo Plugins For a Fixed One-Time Fee
Forums     Plugins     Marquis Compressor Program-Dependent Release Contour

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.

Ok...and finally.  Is there anyway to define to us...in any way.  What makes those filters different from one another.

By the way....THANK YOU for the answering all of these questions.

;)

Elvenking: Ok...so its may be said that, A = One process, B= One Process and C= One Process. (Each being distinct and separate processes) but the combination of them is the system.  Right?

Aleksey Vaneev: Yes, they are.  As I've replied earlier, they are FILTERS.


What makes them different is that you can't represent three filters with a single one.  Other than that they are identical.

Elvenking: Ok...and finally.  Is there anyway to define to us...in any way.  What makes those filters different from one another.


I see.  So all you are saying is, that they are the same filter...but...the sum of them is far more than just an additive effect.  In other words, this explains why we get different results from using a setting of 1.0, 10, 10, than we would with a setting of 10, 10, 1.0.  Okay.  Also, that programatically....filter one is applied first...then filter 2...then filter 3.  It would be the same as applying three filters in series.  So combined, they have some sort of influence over the filter that follows.

Will there be any way to make the graph more adequately reflect the parameters?

Aleksey Vaneev: What makes them different is that you can't represent three filters with a single one.  Other than that they are identical.

Elvenking: Ok...and finally.  Is there anyway to define to us...in any way.  What makes those filters different from one another.


Also, with the default settings of the PDRC...is that the state of the compressor in its most "automatic" state. (Meaning, little or no influence over how Marquis is determining release time)

Aleksey Vaneev: What makes them different is that you can't represent three filters with a single one.  Other than that they are identical.

Elvenking: Ok...and finally.  Is there anyway to define to us...in any way.  What makes those filters different from one another.


Yes, these filters are applied in series.  However, it was not an expected fact that 1,10,10 and 10,10,1 produce different results - probably, this behavior exists due to computation errors only (computers have finite precision, especially with floating point number comparisons).

I do not think I can reflect these parameters more adequately than what we have now.


Elvenking: Also, with the default settings of the PDRC...is that the state of the compressor in its most ''automatic'' state. (Meaning, little or no influence over how Marquis is determining release time)

It is an average setting in my opinion.  PDRC to have little or no influence should be tuned to 0.1,0.1,0.1.


Aleksey, you have done a great job.  Thanks for taking the time to respond today.  It is really appreciated.  No more questions over here.

Aleksey Vaneev: Yes, these filters are applied in series.  However, it was not an expected fact that 1,10,10 and 10,10,1 produce different results - probably, this behavior exists due to computation errors only (computers have finite precision, especially with floating point number comparisons).

Aleksey Vaneev: I do not think I can reflect these parameters more adequately than what we have now.


OK, glad this was helpful.  I'll include some of these details in the user's manual.

I love Elvenking

Ok...posted this on a parallel thread as my best explanation of what the PDRC does.  Aleksey, you can review and correct if you would like:

Ok here it goes, I will do this to the best of my ability.

Firstly, we all know what release time is: The program material (audio) crosses the threshold, activating the gain reduction algorithms, As soon as the audio level falls below the threshold set point, the compressor releases the input signal back to its uncompressed state.  Release time is how much time we give to go from the "reducing gain" state, back to 0 gain reduction.  Ok, basic compressor stuff.

In the case of the Marquis, we have what is called a Program Dependant Release mode.  The states, that there are algorithms inside the code (circutry) that will attempt to analyze the audio transients, and create a release time that is appropriate, with loose reference to the set release time.  So say we have the release time set to 300ms, and we are compressing a cymbal with a long decay.  The compressor may stretch out the release time to accomodate for the extended decay time.  From what Aleksey has told me about the Marquis, this time adjustment should never fall below the set release time.  So no matter how short the transient, the release time should always be at least 300ms.  In essence, if the audio level of the intrument being compressed is falling slowly, the release time is likely to appear longer.

Now, to throw another curve into the equation (pun intended), Aleksey has added the Program Dependant Release Contouring feature.  The thing to recognize here is that this is not an adjustment which directly affects Release Time.  It simply influences the algorithm that Marquis is using to automatically determine release time (the Program Dependant Release Time algorithm discussed above).  So when we adjust this curve, we are adjusting the amount of influence that the audio material has over the calculated release time.  An example might help: Say we still have our example above and we are finding that all of our cymbals through our OH mic, are being released too slow.  We are getting too much of a pumping effect and it seems that things just arent returning to normal level fast enough.  However, if we shorten the release time too much, we are finding that the cymbals with longer decay are being released too fast.  So we like what the Program Dependant Release time is doing for us, but we just want the algorithm to kick back a bit.  So we tweak one of the Parameters (parameters explained soon) from 1.80 to say 3.00 and this tells the PDR Time algorithm to do everything its normally doing...just a little faster.  Cool, we now notice that the Marquis is falling closer to our set Release Time, but still exhibiting desirable analysis of the incoming audio and more mildly adjusting the release time to suit the decays of the various cymbals. (Once again the nature of the original Program Dependant Release Time).  So in a short sentance, the PDRC controls how much the incoming audio and its different transients, affect overall release time.  Cool....I hope we got through that ok.

Finally, the parameters.  They are each the EXACT SAME FILTER. (SO, 10, 10, .1 and .1, 10, 10, and 10, .1, 10 all have the same effect on the PDR Time.  Any differences discovered are purely accidental and likely due to floating point errors.) Each one of them represents a filter that contours the Program Dependant Release Behavior.  The default setting of 1.80, 1.80, 1.80 is where things would be set if there were no controls for PDRC.  It is the default Program Dependant Release Behavior.  If they are all set at the maximum of 10, 10, 10, then the release time will be what you have set the "Release Time" control at.  If they are all cranked down to 0.1, 0.1, 0.1, then the Program Dependant Release Algorithm will be in full effect, and you will unlikely ever see the compressor release the signal back to normal levels.  Gain suppression seems to be perminent with these settings.  You have basically given an exagerrated amount of control over the release time to Marquis. "So Elvenking, why would I adjust Param 1 insted of Param 2...they are the same filter right?" Correct, start with which ever you wish. (Since I am not privilaged to the information pertaining to exactly what the filters do, I can only give a general idea here).  I got the idea from the posts at Voxengo, that one parameter would only be a liner influence over the Program Dependant Release Time.  If you only adjust one slope, you can only have a linear effect over the release algorithm.  However, if you combine three linear filters set at different slopes, you get a curve...correct?  The idea behind it is that, you may get an inital adjustment on one of the filters that you like.  And from there, you simply want to increase the release rate, at the middle of the release time.  So you would tweak another parameter to place a curve onto the existing line (or curve).  And you could tweak another in order to create a sort of S-Curve.(This is all only extrapolated from the information I have, which still has holes.  Aleksey will not give up his trade secrets) The fact that the graph does not adequately depict this, is what makes the whole idea of 3 Parameters seem whacky.  Hopefully the graph will be updated to reflect that sometime.  Most important is the overall curve, but it is also important to realize that the combination of the three Parameters creates a curve, whereas only one, would be a straight line..or liner influence.

CAUTION: Read only if you want to go insane: You can see this by setting two controls for 10, and adjusting the third, you will notice a very linear adjustment taking place as you tweak it.  Now set that control to about 1.0 and adjust one of the knobs set to 10.  You will se a very different adjustment taking place. (Meaning that moving the knob that was set at 10 down by 1 unit had much less of an effect than moving the control set at 1.0 up 1 unit.) WTF did I show you this for?  The parameters are not liner in their adjustments "when combined".

Bottom line, given that the way Marquis automatically determines Release time based on audio input is unknown, and the resulting influence of the PDRC over that algorithm is sort of a shot in the dark, you might as well just begin by knowing that it is the overall curve that is making the biggest difference.  How the "influence curve" is changing otherwise, is anyones guess.  Best to start with the default (1.8, 1.8, 1.8) or the minimum onfulence settings (10, 10, 10)

When looking at the graph, think of the X-axis as release time, and the Y-axis as influence over the Program Dependant Release Time Algorithm.

Ok...spent a long time doing this....time to go get a workout.  I will proof read it later.

This topic was created before release of the latest product version, and it may contain details irrelevant to this version.  Replying is disabled for this topic.