The PAN LAW Discussion... aka What is Cockos doing?

[karbomusic]

Originally Posted by _Stevie_

@Karbo: is there really a need to discuss this before there is a solution to the problem. I mean REALLY?

Oh FFS yes since I asked for a basic yes/no not a discussion; it's worth 10 GD seconds to say what Theonex just said and better than either of you.

[EvilDragon]

Originally Posted by Klangfarben

Also, ErBird should win some kind of medal...

Those formulae don't look very efficient to me.

[karbomusic]

Originally Posted by ErBird

Howdy. I don't have anything more to add. I feel like I demonstrated the problem and offered a solution. So my work is done. Not looking for glory, just what's right.

Of course if the math is different so will be the track output in some cases. But is that a good enough reason to avoid fixing what is broken? IDK. It's just a bummer it was broken to begin with.

Theonex covered the main thing I was wondering but thanks!

[ErBird]

.....

[Klangfarben]

Originally Posted by EvilDragon

Those formulae don't look very efficient to me.

What specifically do you think doesn't look efficient about them? (edit: not being snarky here. Genuinely curious)

[pepe44]

Originally Posted by schwa

Looks like Nuendo defaults to a linear taper, which gives the first stereo image I posted above. REAPER defaults to a sin taper which gives the second image. This is why REAPER offers 5 different modes. In any case, this is not a bug.

and...

Originally Posted by Justin

There is not really a correct default, as REAPER’s tracks are multi-purpose. Folder tracks, bus tracks, tracks with stereo media, and tracks with mono media would all benefit from different pan law configurations...

0dB being the default law gives predictability with gain, and if you start panning you can tweak that track as desired. Or if the use of the track is known before hand, it can be tweaked in a template...

This is your right answer.

[Thonex]

Originally Posted by pepe44

I support you on the label.

Make sure you didn't miss this bit:

https://forum.cockos.com/showpost.ph...3&postcount=61

[Klangfarben]

Originally Posted by Justin

There is not really a correct default, as REAPER’s tracks are multi-purpose. Folder tracks, bus tracks, tracks with stereo media, and tracks with mono media would all benefit from different pan law configurations...

0dB being the default law gives predictability with gain, and if you start panning you can tweak that track as desired. Or if the use of the track is known before hand, it can be tweaked in a template...

Originally Posted by pepe44

This is your right answer.

Justin is simply explaining why they chose 0dB as the default pan setting. Not that there is actually something wrong in the math, which I think has been pretty conclusively proven at this point unless someone is finding errors in ErBird's data.

[schwa]

Does anybody want to set up cubase/nuendo with a +0dB pan law, stereo tone signal, automate the pan to a full left/right sweep, and put a stereo image visualizer plugin on it (like schope in phase mode)? I suspect that this is what you'll see.

[Thonex]

Originally Posted by schwa

Does anybody want to set up cubase/nuendo with a +0dB pan law, stereo tone signal, automate the pan to a full left/right sweep, and put a stereo image visualizer plugin on it (like schope in phase mode)? I suspect that this is what you'll see.

I'm happy to help. I'll do it right now.

With regards to visualizers... let me see what I can do...

Stay tuned.

[Thonex]

Ok... here is a test... using Izotope 7's Imager.

Nuendo's 0dB Pan Law

There is no freeze function so I made a gif. I hope this helps. Maybe there is a demo version of that schope I can try?

[Thonex]

Ahh,... saw you gave a link and there is a free trial... give me a few minutes

[ashcat_lt]

Originally Posted by schwa

Does anybody want to set up cubase/nuendo...

Why the hell does it matter?

There should never be gain with 0db Pan Law. Period.

Likewise, as you pan away from a channel in any Pan Law, that channel should never be quieter than at center. Period.

[Thonex]

Here it i with the Schope.

I made a gif again since I wasn't able to grab a script shot that look clean

Interestingly... in Nuedno, it looked 100% the same at 0dB as -3dB Pan Law.



I hope this helps.

[ErBird]

Originally Posted by Thonex

I made a gif again since I wasn't able to grab a script shot that look clean

I think the level is too high for the "Polar Sample" meter to see the top. You may need to turn it down.

[Thonex]

Originally Posted by ErBird

I think the level is exceeding the meter. You may need to turn it down.

It's a -12dB tone.

Don't be confused by the slider marking. The VU window numbers are what you should be looking at.

[ErBird]

Anyway, I had to test efficiency after someone brought it up. This is un-optimized, exactly as written in Desmos. Hopefully no bugs.

desc:Sine Panner

slider1:0<-100,100,1>Pan
slider2:0<0,4,1{0.0 dB,-2.5 dB,-3.0 dB,-4.5 dB,-6.0 dB}>Pan Law

@slider

p = 0.005*slider1 + 0.5;

slider2 == 0 ? // 0.0 dB
(
  0 <= p && p <= 0.5 ?
  (
    pan_l = 1;
    pan_r = 2*p;
  ):(
    pan_l = 2*(1-p);
    pan_r = 1;
  );
);

slider2 == 1 ? // -2.5 dB
(
  pan_l = sin((0.824990077532479)*sin(($pi/2)*(1-p)))/sin(0.824990077532479);
  pan_r = sin((0.824990077532479)*sin(($pi/2)*p))/sin(0.824990077532479);
);

slider2 == 2 ? // -3.0 dB
(
  pan_l = sin((0.11924678268314)*sin(($pi/2)*(1-p)))/sin(0.11924678268314);
  pan_r = sin((0.11924678268314)*sin(($pi/2)*p))/sin(0.11924678268314);
);

slider2 == 3 ? // -4.5 dB
(
  pan_l = sin(($pi/2)*0.731621621646835*(1-p))/sin(($pi/2)*0.731621621646835);
  pan_r = sin(($pi/2)*0.731621621646835*p)/sin(($pi/2)*0.731621621646835);
);

slider2 == 4 ? // -6.0 dB
(
  pan_l = sin(($pi/2)*0.0876554612819604*(1-p))/sin(($pi/2)*0.0876554612819604);
  pan_r = sin(($pi/2)*0.0876554612819604*p)/sin(($pi/2)*0.0876554612819604);
);

@sample

spl0 *= pan_l;
spl1 *= pan_r;

0.0 dB compared to Reaper


The other pan laws are fairly close in Reaper, with -2.5 dB being the only other one that boosts the signal.

[schwa]

Thanks, that's what I expected to see.

With the caveat that the perceived loudness of a combined signal depends on the frequency content of the material (which can maybe help put this conversation in some perspective, too) --

The linear stereo images shown in posts 89 (REAPER with "3.x balance" pan mode) and 94 (Nuendo) and 97 (erbird's plugin, the left image) are the only shape you can create for any pan law above -3dB without boosting one channel. With any higher pan law, the only way to create a sinusoidal image is to boost one channel.

So if you want the linear image that has a sharper perceived level difference between center and 50% pan either side, you can use the 3.x mode, and if you want the broader image with less perceived level difference, you can use the default mode.

(Because of the frequency dependence, with the 3.x mode, high frequency content may sound quieter at 50% pan than when centered. With the default mode, low frequency content may sound louder at 50% pan than when centered.)

As Thonex pointed out, for some particular use cases, there might be other reasons why the channel boosting is particularly undesirable, so you can use the 3.x mode in that case.

[Thonex]

Originally Posted by schwa

As Thonex pointed out, for some particular use cases, there might be other reasons why the channel boosting is particularly undesirable, so you can use the 3.x mode in that case.

Thanks as always schwa!

I would submit that most users don't put on a lab coat like me and maybe the "deprecated" one should not be so "deprecated" since it would seem the rest of the DAW industry has adopted this type of Pan Law. Users just want it to work like everywhere else.

This reared it's ugly head with the gain staging anomalies (which I posted) which led me to the Pan Laws... none of which satisfied me except the deprecated one. I'll leave it in Cockos' capable hands to decide what to do, but I'd argue the current Reaper default behavior is not in line with the rest of the industry.

Thanks so much for your attention on this topic devs!!

[EvilDragon]

As if Reaper always intended to be in line with the rest of the industry.

[schwa]

If the conclusion of this whole thread is that we should rename the 3.x mode “linear image” or something, that would be just fine.

[EvilDragon]

I think what's tripping Thonex up more is that setting 3.x deprecated to 0 dB has the same behavior as -3 dB equal power pan law in Cubase...

But considering his GIFs from Cubase show stereo meters, implying a stereo track, this means pan law wasn't used, but just the usual stereo balancer that has NOTHING to do with pan laws? That's what's tripping me.

[ashcat_lt]

Originally Posted by Thonex

...but I'd argue the current Reaper default behavior is not in line with the rest of the industry.

Makes exactly zero sense and we're still only hearing lame excuses, but I understand you're trying to be gracious.

[Thonex]

Originally Posted by ashcat_lt

Makes exactly zero sense and we're still only hearing lame excuses, but I understand you're trying to be gracious.

Yeah... we're all on the same team. We all want Reaper to Rock!! And it does!! Except... when it doesn't 😀... and then Thank God the devs are so responsive. We need to appreciate this... this is a unique Cockos thing where development interaction with the user base is unlike anything I've ever seen.

[EvilDragon]

BTW Thonex, check this thread out.

https://forum.cockos.com/showthread.php?t=42821

2009. Back then people were saying Reaper's pans are wrong, because they act as a balancer (which is the exact same thing you showed Cubase was doing! But this is not pan law since you're showing a stereo track, so of course it will act as a balancer rather than apply the pan law...). So they added new tapers that l0calh05t made in that thread.

And now 11 years later they're apparently wrong again? Irony is super hard on this one. At least you have options!

[Thonex]

Originally Posted by EvilDragon

I think what's tripping Thonex up more is that setting 3.x deprecated to 0 dB has the same behavior as -3 dB equal power pan law in Cubase...

But considering his GIFs from Cubase show stereo meters, implying a stereo track, this means pan law wasn't used, but just the usual stereo balancer that has NOTHING to do with pan laws? That's what's tripping me.

Schwa asked for stereo... I gave him stereo. And as mentioned various times... MONO and Stereo gave the same exact results. Test it yourself

[EvilDragon]

I don't have Cubase or Nuendo to test it out. Nor Logic, nor PT.

But I'm implying to post #13, where you show Nuendo for the first time, and saying that's -3 dB pan law, but pan law is not being used there at all, it's just a stereo balancer. Unless I am missing something there.

[Judders]

Originally Posted by EvilDragon

I don't have Cubase or Nuendo to test it out. Nor Logic, nor PT.

But I'm implying to post #13, where you show Nuendo for the first time, and saying that's -3 dB pan law, but pan law is not being used there at all, it's just a stereo balancer.

You do have Kontakt, so you can see for yourself that it a stereo balancer.

[EvilDragon]

Yeah it's a stereo balancer. No pan laws there. Or am I missing something?

[Thonex]

Originally Posted by schwa

If the conclusion of this whole thread is that we should rename the 3.x mode “linear image” or something, that would be just fine.

I think "Linear Image" would mean absolutely nothing to anyone except those of us that have read this thread.

I'm also unsure bout the "0dB" labeling when it behave like -3dB on other DAWs. But maybe putting at the top of the list and not naming it "deprecated" would be helpful?

Maybe "Common Pan Law (like most other DAWs)" or something more user friendly. But I would check with PT to make sure the its not just Cubendo, Logic and Kontakt.

[Thonex]

Originally Posted by EvilDragon

Yeah it's a stereo balancer. No pan laws there. Or am I missing something?

I think what you're missing is the unwanted gain-staging

Ok... back to work for me.

Thanks everyone for shedding light on this.

[EvilDragon]

PT uses -2.5 dB pan law.

[EvilDragon]

Originally Posted by Thonex

I think what you're missing is the unwanted gain-staging

I didn't really need that somewhat snarky response, it was an honest question

It's just: stereo balancer, no pan laws are applied. Stuff is linear. You turn one channel down while the other stays where it is, until they flip over.


As for 0 dB vs -3 dB thing, this quote from a Sound On Sound article might shed some light... "Some mixer and DAW pan controls work by attenuating the level of both left and right output channels as the pan control is moved towards the centre. Others work by raising the level of one channel as the pan moves towards that side."

[ErBird]

Originally Posted by EvilDragon

And now 11 years later they're apparently wrong again? Irony is super hard on this one.

Yes. They are.

[Thonex]

Originally Posted by EvilDragon

I didn't really need that somewhat snarky response, it was an honest question

It's just: stereo balancer, no pan laws are applied. Stuff is linear. You turn one channel down while the other stays where it is, until they flip over.


As for 0 dB vs -3 dB thing, this quote from a Sound On Sound article might shed some light... "Some mixer and DAW pan controls work by attenuating the level of both left and right output channels as the pan control is moved towards the centre. Others work by raising the level of one channel as the pan moves towards that side."

I know about the center panning reduction vs LR gain to compensate. these would not have led to what I was documenting. Also, no offense.. just having fun with you. Sorry for being snarky. I guess I don't really care how each DAW or KT does when it's predictable and doesn't result in gain-stage problems.

Ok... but back to work for real!

[karbomusic]

Originally Posted by EvilDragon

it was an honest question

I think you'll be better up to speed if you revisit ashcat's responses - best I can tell that changing of level when it shouldn't is where the meat is but not sure - like I said doesn't affect me much.

[Heart Doctor]

Originally Posted by Thonex

Here it i with the Schope.

I made a gif again since I wasn't able to grab a script shot that look clean

Interestingly... in Nuedno, it looked 100% the same at 0dB as -3dB Pan Law. ...

It could be that your tone generator feeds a stereo track instead of a mono track.
In this case you always have a 0dB Stereo Balance, no matter how the pan law
is set in the Cubase project settings.

But back to the main topic:

I compared the stereo pan settings of REAPER one more time with Cubase.

As a result, I now can confirm the observations of Thonex and ErBird:

The behavior of REAPER's "Stereo balance / mono pan" only matches Cubase
(and possibly other DAWs) when gain is set to -3dB, -4.5dB or -6dB.
But not for 0dB.

REAPER's paning behavior for 0dB gain is unusual and not at all what we
would expect from other gain settings of the same panning curve.
The 0dB behavior now looks broken to me too.

In case of the deprecated 3.x balance, 0dB behavior is fine and consistent
with other gain settings of the same panning curve.

By the way, the "Equal Power" pan law in Cubase is a -3dB gain pan law.
Equal Power generates a circular image in a stereo scope (lower picture
posted by Schwa) and corresponds to REAPER's current -3dB pan law.

All other pan law settings in Cubase result in linear taper, i.e. triangular
images in the stereo scope. Similar to REAPER's deprecated pan law.

[ashcat_lt]

Originally Posted by ErBird

Just like the "weird knee" in ReaComp which, just like in this case, was fixed improperly.

It actually LOOKS a lot like that problem with ReaComp folding over at the knee, but I think that's a bit of a different mechanism. You've got me curious though. What do you think is wrong with the new behavior? It doesn't fold over anymore, and I'm quite happy with the nice predictable curve that it's got now.

[ashcat_lt]

Originally Posted by karbomusic

Either way, I'm sure someone will answer. Ashcat?

You already got your answer on this, but... If it was just plain changed, it would technically "break" backward compatibility, at least for certain things panned certain ways, but the difference is tenths of db at worst, so nobody would probably notice unless they tried a null test.

[karbomusic]

Originally Posted by ashcat_lt

You already got your answer on this, but... If it was just plain changed, it would technically "break" backward compatibility, at least for certain things panned certain ways, but the difference is tenths of db at worst, so nobody would probably notice unless they tried a null test.

Thanks man.