Is this really how Pro Tools does panning? Take the test

[rhythminmind]

Panning ITB does not = Less fidelity for digital masters. It would only hold true if your final mix was to an analog medium.

Hypothetical Example.

ITB
Stereo Track one has a left heavy pan.
The Left channel is using all 24 bits of headroom.
The right channel is say using 12 bits of headroom do to the lower level.
Final ITB 24 bit mix-down = Left 24 - Right 12

OTB
Stereo Track one panned center.
Both Left & Right channels are using all 24 bits of headroom out of the D/A.
You make the same left heavy pan OTB.
You now have to print the mix back into the DAW.
The Left channel is using all 24 bits of A/D headroom.
The Right channel is using 12 bits of A/D headroom do to the lower level.
Final 24 bit mix = Left 24 - Right 12 (But now with a noizey analog noise floor )

I kept the numbers simple for easy reading.
FYI - I am by no means a ITB advocate, I actually mix OTB. But for processing & workflow reasons. Not for fidelity.

[monomer]

Originally Posted by Clueless

Yes, if you think that 144dB is enough resolution.

Exactly my point, and with 48 bits it sould be waaaay higher than that.

Or, let's do the actual research and discover that ProTools has a 48 bit mix bus.

I just assumed it cause i know that it will be at least 32bits.
More is of course better, but 32bits should already have you out of any panning problem territory.

BTW, the 48 bit Pro Tools mix bus is not necessarily the be-all end-all in digital resolution: the Harrison mix bus is 64 bit floating point, which provides 48 bit resolution across a virtual infinity of levels. Thus, if you were mixing two teeny-tiny signals in a very quiet section, they'd preserve all their resolution, just as when mixing two REALLY LOUD signals.

But then what is the use of mixing stuff with a dynamic range that exceeds the human pain threshold?

But back to ProTools, the math that they publish make it pretty clear that they have their math right. Whether the math is as good as analog is the good old analog v. digital question, but their math is not out-and-out stupid.

Digital summing is no rocket science. Sure, you need to take into account some things about how computers deal with numbers, but appart from that its just about adding stuff.

[okydoky]

Originally Posted by rhythminmind

Panning ITB does not = Less fidelity for digital masters. It would only hold true if your final mix was to an analog medium.

Hypothetical Example.

ITB
Stereo Track one has a left heavy pan.
The Left channel is using all 24 bits of headroom.
The right channel is say using 12 bits of headroom do to the lower level.
Final ITB 24 bit mix-down = Left 24 - Right 12

OTB
Stereo Track one panned center.
Both Left & Right channels are using all 24 bits of headroom out of the D/A.
You make the same left heavy pan OTB.
You now have to print the mix back into the DAW.
The Left channel is using all 24 bits of A/D headroom.
The Right channel is using 12 bits of A/D headroom do to the lower level.
Final 24 bit mix = Left 24 - Right 12 (But now with a noizey analog noise floor )

I kept the numbers simple for easy reading.
FYI - I am by no means a ITB advocate, I actually mix OTB. But for processing & workflow reasons. Not for fidelity.

hi,

your theory does not appear entirely correct.

you are not taking into account the difference between what happens when pro tools sends a signal to its outputs and what happens when a signal is recorded directly into the adc. for example there is dither that helps and you can raise the level of the signal before recording. almost all modern converters are oversampling converters, employing a whole bunch of interpolation and averaging and so forth.

you could record to dsd or something, if you are real concerned.

also, under your theory, by doing the panning in pro tools and then re-recording it back into pro tools, rather than panning it in the analog domain and then re-recording it, you are creating "the problem" twice [if you believe that resampling the mixed signal results in the same issue, which i do not believe to be the case].

one trouble with "in the box" is the fact that it is not practical for lots of material. so people end up being "almost in the box", which blows it.

theories of maintaining the "digital integrity" of the signal by never having to go in and out of conversion fall apart if you have to do that at all. and people generally do.

so what happens is that people start making processing decisions based not on what would sound best, but, rather, based on what will allow them to stay in the box.

i see a lot of people posting where you can tell they have this kind of obsession with making it all work inside their computer. i experienced that myself in trying avoid any further a to do conversion beyond the initial tracking. almost impossible to do unless you severely limit your options.

working completely in the box is suitable for some types of projects, and you can get a convincing result that way if you stay within the limits of what it has to offer. i can record and mix "in the box" just fine, but i do not really see the point in making it some sort of goal, particularly with all that analog recording has to offer.

most people that really have the option to use "high end" outboard gear and a good console will choose to do that rather than trying to accomplish the artistic goals using only plug-ins. the advantages of being able to use any of the amazing new or vintage analog outboard devices is formidable, and it does not appear that plug-ins have made that stuff obsolete by any means.

so this discussion is not about the never-ending non-controversy over "in the box" vs. "out of the box". we can assume at this point that at least some, if not most, people will work "out of the box", at least some of the time, either by choice, or at someone else's direction, or by necessity.

therefore, if you are using pro tools "out of the box", you might as well do the panning out of the box to avoid some of the limitations.

the fact that you are re-sampling the analog mix back into pro tools [or some other device] does not negate the benefit, because it does not create the very same problem all over again.

furthermore, staying in the box also does not eliminate "the panning problem" because in order to stay in the box you are going to have to pull some channel faders way back, and if you are panning there the "opposite channel" may very well go below the [alleged] 90dB leeway.



right.

[SeanG]

I can hear 100 different pan locations in Nuendo.
I don't use Pro Tools so I cannot speak to its panning issue.

Itb panning here sounds wider and more accurate than otb. Maybe do to the lack of crosstalk.

However I do use a hybrid system for analog summing, which is where I hear an improvement over itb. This stage obviously would have some crosstalk, essentially diluting the accuracy of itb panning. This is a short coming that I'm willing to except due to the summing improvements that I hear and like.

As far as truncating one side or the other isn't it a moot point? How many times do you have all audio coming from one side and nothing on the other?... For me never. Even when I have an intro or a break down that has one hard panned element (which is rare), I still usually have enough stereo delay or reverb that clears any very low dbfs level that would cause truncation on the other channel.

Whatever the case I'm totally happy with panning itb and summing otb. I hear no shortcomings in the panning of Nuendo and I like the recall and automation that itb provides. And I'm totally happy summing otb, my large track count songs sound punchier and have more depth. For me this is what I like, if someone else likes their methodology better, and gets the results that they're listening for than I'm cool with that.

If Pro Tools panning is as bad as some of you are saying, I would also be disappointed and would definitely move to panning otb. 5 panning locations really? I’ve been considering adding a Pro Tools rig to my set-up, so I'll have to do some more research.

[okydoky]

Originally Posted by SeanG

I don't use Pro Tools so I cannot speak to its panning issue.


As far as truncating one side or the other isn't it a moot point? How many times do you have all audio coming from one side and nothing on the other?... For me never. Even when I have an intro or a break down that has one hard panned element (which is rare), I still usually have enough stereo delay or reverb that clears any very low dbfs level that would cause truncation on the other channel.

hi,

it does not work that way. each channel is on a separate output, or outputs [generally, if you are mixing out of the box into a console].

also, low level components of sounds [like harmonics] can be subject to degradation due to quantization noise even if the overall signal volume of the channel is higher due to other elements.

that's why, for example, low bit depth recordings of musical instruments do not sound as life-like.

right.

[okydoky]

hi,

does anyone know if digidesign recommends use of the dithered mixer when using multiple outputs out of the box, or only when mixing in the box to a stereo out?


right.

[Nishmaster]

Lots of misinformation in this thread. Oky**** is simply not correct. Panning ITB is not less accurate than OTB. I would say the opposite, really; the tracking of even high quality dual-log pots is not perfect. If the panning is VCA based, well then you are running your signal through a VCA which can be (relatively speaking) a high distortion trip. But all the faults of analog are irrelevant here. Use what tools you like using.

However, the notion that digital panning is inferior because of increased quantization distortion due to a reduction in level and thus bit depth is completely false. In a dithered system, decreasing bit depth DOES NOT increase quantization distortion, it ONLY increases the noise floor. Quantization distortion is eliminated by the dither and replaced with a (marginally) higher noise floor. The effects of low level panning on digital and analog are thus identical.

Let us take a 24 bit system with a noise floor of -144dB. Let us also consider a modern console with a noise floor of around -110dB (really good, even by modern standards). Your dither, toggling the LSB, will increase the noise floor by 6dB, but in doing so, will completely and totally ELIMINATE quantization distortion. So now, you are left with essentially two identically behaving devices (as far as the constraints of this thought experiment, noise floor, are concerned). Let us say that the analog system and digital system are calibrated so their dB readings match. Putting a 0dB signal through the system (and ignoring pan law):

Digital Box:
100 | 0 = 0db L | -inf R
90 | 10 = -14.4dB L | -129.6dB R
In the digital box, we are a good deal above the noise floor here.

Analog Box:
100 | 0 = 0db L | -inf R (ignoring crosstalk)
90 | 10 = -14.4dB L | -129.6dB R
In the analog box, this pan is still 19.6dB under the noise floor.

Which of these seems like the more optimal situation? Again, my intent is not to spark yet another endless ITB vs. OTB debate, but to debunk this particular inaccurate bit of misinformation.

[okydoky]

Originally Posted by Nishmaster

Lots of misinformation in this thread. Oky**** is simply not correct. Panning ITB is not less accurate than OTB. I would say the opposite, really; the tracking of even high quality dual-log pots is not perfect. If the panning is VCA based, well then you are running your signal through a VCA which can be (relatively speaking) a high distortion trip. But all the faults of analog are irrelevant here. Use what tools you like using.

However, the notion that digital panning is inferior because of increased quantization distortion due to a reduction in level and thus bit depth is completely false. In a dithered system, decreasing bit depth DOES NOT increase quantization distortion, it ONLY increases the noise floor. Quantization distortion is eliminated by the dither and replaced with a (marginally) higher noise floor. The effects of low level panning on digital and analog are thus identical.

hi,

first off, i am not the poster who was talking about the pan law stuff, or the "number of pan points in pro tools vs analog", so i'll let you address those issues with them. i think people generally agree that panning is smoother and less clunky when you do it on a console. maybe if you are using one of digi's control surface thingys it works better, i don't know.

however, your basic dither analysis is erroenous here insofar as it is not relevant to the particular discussion. that is simply because we are not talking about "in the box 48-bit mixer" dither.

i have posted the following twice already in this thread, but maybe you did not see it.

48-bit mixer, at page 6.

"We should take a moment to distinguish between the result of lowering a fader within the context of the internal mixer (where 48-bit precision is maintained) and lowering the level of a signal that is being sent to a DAC output.

As with any digital audio system (not just Pro Tools), if you
lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."


so we are talkiing about a scenario where you are not just using a stereo output pair to you monitors, but, rather a complete system with several interfaces, with say 48 separate outputs to a console. again, according to digidesign, l

"if you lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly".

so your argument may be with digidesign [if you say that their statement is wrong], and not with me.

the other issue that i was trying to hint at above, is that digidesign does not really recommend the use of the dithered mixer in using multiple outputs. to wit [also from the "48-bit mixer paper"]:

"Dither is not added on a track by track basis, as this would produce unwanted accumulated noise. Dither is only added once at the Master Fader output of the mix bus so the total system dither noise is 3 dB at -144 dB"..


right.

[Etch-A-Sketch]

Originally Posted by Nishmaster

Lots of misinformation in this thread. Oky**** is simply not correct. Panning ITB is not less accurate than OTB. I would say the opposite, really; the tracking of even high quality dual-log pots is not perfect. If the panning is VCA based, well then you are running your signal through a VCA which can be (relatively speaking) a high distortion trip. But all the faults of analog are irrelevant here. Use what tools you like using.

However, the notion that digital panning is inferior because of increased quantization distortion due to a reduction in level and thus bit depth is completely false. In a dithered system, decreasing bit depth DOES NOT increase quantization distortion, it ONLY increases the noise floor. Quantization distortion is eliminated by the dither and replaced with a (marginally) higher noise floor. The effects of low level panning on digital and analog are thus identical.

Let us take a 24 bit system with a noise floor of -144dB. Let us also consider a modern console with a noise floor of around -110dB (really good, even by modern standards). Your dither, toggling the LSB, will increase the noise floor by 6dB, but in doing so, will completely and totally ELIMINATE quantization distortion. So now, you are left with essentially two identically behaving devices (as far as the constraints of this thought experiment, noise floor, are concerned). Let us say that the analog system and digital system are calibrated so their dB readings match. Putting a 0dB signal through the system (and ignoring pan law):

Digital Box:
100 | 0 = 0db L | -inf R
90 | 10 = -14.4dB L | -129.6dB R
In the digital box, we are a good deal above the noise floor here.

Analog Box:
100 | 0 = 0db L | -inf R (ignoring crosstalk)
90 | 10 = -14.4dB L | -129.6dB R
In the analog box, this pan is still 19.6dB under the noise floor.

Which of these seems like the more optimal situation? Again, my intent is not to spark yet another endless ITB vs. OTB debate, but to debunk this particular inaccurate bit of misinformation.

I agree.

one other thing you need to do in Oky****'s test to hear the quantization noise (as he mentioned) is you have to turn down/off the channel that the signal is panned to and then turn UP the side where there is no signal. This is by far a "real world" scenario.

First, as stated, dither will remove quantization noise. Second, this quantization noise is happening down around -130dB ~ -144dB... The noise floor for the BEST converters are -120dBfs... So right there, the quantization noise is left undithered would fall below the noise of the DAC. And then there is the noise of the analog console, As stated the noise floor in an analog console is around -90dB~ -110dB, which is even louder than the noise floor in the DAC. In a properly calibrated and dithered system, you will never hear any of the artifacts described because it falls below the noise floor of the system...the only way to hear it is to run the system outside of it's intended specs. Which is what happens when you, for example, use an undithered mixer and pan something hard left, then pull the left channel down on the console and turn the right channel up in order to get the undithered quantization noise above the noise floor of the console. It's pointing out a problem that isn't really a problem because you will never hear it if your system is calibrated and dithered properly.

Anyway...that's my $0.02 I'm not an EE or CS, nor do I design DAWs... but in my experiences that's what I have found. These claims are only founded if you run your system in a way it wasn't intended to be run.

[okydoky]

Originally Posted by Nishmaster

Lots of misinformation

Digital Box:
100 | 0 = 0db L | -inf R
90 | 10 = -14.4dB L | -129.6dB R
In the digital box, we are a good deal above the noise floor here.

Analog Box:
100 | 0 = 0db L | -inf R (ignoring crosstalk)
90 | 10 = -14.4dB L | -129.6dB R
In the analog box, this pan is still 19.6dB under the noise floor.

Which of these seems like the more optimal situation? Again, my intent is not to spark yet another endless ITB vs. OTB debate, but to debunk this particular inaccurate bit of misinformation.

hi,

also, -129.6 dB R is not "a good deal above the noise floor". the theoretical 144dB is the entire range for 24 bit audio, and the bottom part of that range suffers. it is not as if nothing bad happens until after you reach or exceed 144dB down. you can't exceed -144dB down, and the quality of the audio begins to degrade before that.

further, you are comparing analog noise to quanitzation noise, which is altogether improper. you can hear signal well down into an analog noise floor. you cannot do so with digital noise.


right.

[okydoky]

Originally Posted by Etch-A-Sketch

I agree.

one other thing you need to do in Oky****'s test to hear the quantization noise (as he mentioned) is you have to turn down/off the channel that the signal is panned to and then turn UP the side where there is no signal. This is by far a "real world" scenario.

First, as stated, dither will remove quantization noise. Second, this quantization noise is happening down around -130dB ~ -144dB... The noise floor for the BEST converters are -120dBfs... So right there, the quantization noise is left undithered would fall below the noise of the DAC. And then there is the noise of the analog console, As stated the noise floor in an analog console is around -90dB~ -110dB, which is even louder than the noise floor in the DAC. In a properly calibrated and dithered system, you will never hear any of the artifacts described because it falls below the noise floor of the system...the only way to hear it is to run the system outside of it's intended specs. Which is what happens when you, for example, use an undithered mixer and pan something hard left, then pull the left channel down on the console and turn the right channel up in order to get the undithered quantization noise above the noise floor of the console. It's pointing out a problem that isn't really a problem because you will never hear it if your system is calibrated and dithered properly.

Anyway...that's my $0.02 I'm not an EE or CS, nor do I design DAWs... but in my experiences that's what I have found. These claims are only founded if you run your system in a way it wasn't intended to be run.

i have posted the following three times already in this thread, but maybe you did not see it.

48-bit mixer, at page 6.

"We should take a moment to distinguish between the result of lowering a fader within the context of the internal mixer (where 48-bit precision is maintained) and lowering the level of a signal that is being sent to a DAC output.

As with any digital audio system (not just Pro Tools), if you
lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."


so we are talkiing about a scenario where you are not just using a stereo output pair to you monitors, but, rather a complete system with several interfaces, with say 48 separate outputs to a console. again, according to digidesign, l

"if you lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly".

so your argument may be with digidesign [if you say that their statement is wrong], and not with me.

the other issue that i was trying to hint at above, is that digidesign does not really recommend the use of the dithered mixer in using multiple outputs. to wit [also from the "48-bit mixer paper"]:

"Dither is not added on a track by track basis, as this would produce unwanted accumulated noise. Dither is only added once at the Master Fader output of the mix bus so the total system dither noise is 3 dB at -144 dB"..


right.

[Nishmaster]

Originally Posted by oky****

...so we are talkiing about a scenario where you are not just using a stereo output pair to you monitors, but, rather a complete system with several interfaces, with say 48 separate outputs to a console...

Ok. I may have misunderstood your argument.

However, in the case you describe above, there is no good reason to utilize the DAWs panning and level systems. In this case you would simply be using your analog console's pan and fader.

If you are running a limited amount of stems, by all means, utilize your dither properly and you will eliminate any quantization distortion. Note that noise does strictly add (ie 3dB noise + 3dB noise does not create 6dB noise).

also, -129.6 dB R is not "a good deal above the noise floor". the theoretical 144dB is the entire range for 24 bit audio, and the bottom part of that range suffers. it is not as if nothing bad happens until after you reach or exceed 144dB down. you can't exceed -144dB down, and the quality of the audio begins to degrade before that.

further, you are comparing analog noise to quanitzation noise, which is altogether improper. you can hear signal well down into an analog noise floor. you cannot do so with digital noise.

The bottom part of that range does NOT suffer. I would recommend reading some posts by Paul Frindle and Dan Lavry on the subject. Once dither has occurred, analog and quantization noise are identical. Dither erases quantization noise and replaces it with (slightly elevated) white noise. Distortion of the waveform does not increase as level decreases in any way.

[okydoky]

Originally Posted by Nishmaster

Ok. I may have misunderstood your argument.

However, in the case you describe above, there is no good reason to utilize the DAWs panning and level systems. In this case you would simply be using your analog console's pan and fader.

If you are running a limited amount of stems, by all means, utilize your dither properly and you will eliminate any quantization distortion. Note that noise does strictly add (ie 3dB noise + 3dB noise does not create 6dB noise).



The bottom part of that range does NOT suffer. I would recommend reading some posts by Paul Frindle and Dan Lavry on the subject. Once dither has occurred, analog and quantization noise are identical. Dither erases quantization noise and replaces it with (slightly elevated) white noise. Distortion of the waveform does not increase as level decreases in any way.

hi,

however, again, as stated in the exerpt from digidesign's paper that i posted a few times [i won't post it again here] any signal routed to an output [1 output or many outputs] will not have the dither benefit associated with "48 bit dithered mixer". so, unless you are literally mixing "in the box" [and by that i mean not using any outside summing, but rather bouncing your mix directly to disk within the computer], the exception for signals routed to an output applies.

the argument about the 48-bit dithered mixer eliminating quantization noise [and i am aware of that] is not relevant, or effective, where you are routing channels to outputs and not bouncing directly to disk.

so any channel that you have assigned to an output at all will be subject to the exception talked about in digidesign's paper. that means that even if you have a "master fader" also present in the software mixer, that is controlling a bunch of channels that are routed to outputs 1 -2, those channels will be subject to the exception.

we know that a master fader is present, if invisible, at every output pair, and adding a master fader in the mixer just places a graphic handle in the window.

perhaps you could route a bunch of channels to an aux [or auxes] rather than to outputs, do some software panning on the individual channels, then assign the aux[es] to outputs 1 and 2. that may be a way to invoke the dithered mixer for panning and still route to physical outputs to mix.

i do not see getting more than a very few pairs of dithered channels going at once. we saw that problem years ago with the apogee stuff. people recording every channel 20 bits with uv22 down to 16 into pro tools, and then combining them, rather than just applying it once at the last step. i think digidesign realizes this, and is not recommending it.

different opinions about dither use abound, as i am sure you are aware.

obviously, much of this is subject to the demands of the mix. many mixes will not even have any element that is not either hard left, hard right, or center. but some do. film mixes can get huge and there are all kinds of panning nuances in some stuff.

i reject the argument someone was making to the effect that low level components in the mix are not significant or sufficiently perceivable to warrant attention to fidelity.



right.

[Benmrx]

This could be my mind playing tricks on me, but in PTLE I feel like I have a much, much easier time panning when I'm mixing into a buss with a multi-mono plug like AC1 on it with the L+R sides unlinked.

I've tried all L+C+R panning, and I just can't do it. Everything seems so distracting to me coming from so far left and right all the time, and the center channel get's pretty mushy. I'm not saying it can't be done, and I'm sure plenty of amazing mixes have been done that way, but it's just not for me.

Panning is a pretty huge deal IMO, and only having 3,5,7 or what ever points to use seems so restricting. Just my .02.

That being said, I do wish I could go in and change the panning law in PTLE to -3.

[Etch-A-Sketch]

Originally Posted by oky****

the argument about the 48-bit dithered mixer eliminating quantization noise [and i am aware of that] is not relevant, or effective, where you are routing channels to outputs and not bouncing directly to disk.

so any channel that you have assigned to an output at all will be subject to the exception talked about in digidesign's paper. that means that even if you have a "master fader" also present in the software mixer, that is controlling a bunch of channels that are routed to outputs 1 -2, those channels will be subject to the exception.

we know that a master fader is present, if invisible, at every output pair, and adding a master fader in the mixer just places a graphic handle in the window.

perhaps you could route a bunch of channels to an aux [or auxes] rather than to outputs, do some software panning on the individual channels, then assign the aux[es] to outputs 1 and 2. that may be a way to invoke the dithered mixer for panning and still route to physical outputs to mix.

I'm not sure what you are trying to say here, I don't think you fully understand how the dithered mixer works... Any channel assigned to ANY output has 24bit dither added to it, regardless of whether it is an internal bus or an actual physical output. When anything is assigned to a bus or an output, a 48bit mixer gets created for it. That 48bit mixer outputs a 24bit signal regardless of whether it's internal or external. And when using the dithered mixer, that 24bit output has 24bit dither added to it.

Routing channels to an aux for panning is no different than routing channels to a physical output. When using the dithered mixer, all 48bit mixers get 24bit dither added to them on output.

From your last post it sounds like you are assuming that the audio within PT HD is 48bit. It isn't, only the mixer is. The TDMII architecture is only 24bits wide and the TDM chips can only process 24bit words. So, while the mixer is 48bit (56 with the 8bit accumulator), audio from disk, audio passed from plugin to plugin, audio outputted to a bus or hardware output, and audio received at the input of an audio track or aux (from either a hardware input or a bus) is ALWAYS 24bit.

[Etch-A-Sketch]

Originally Posted by Benmrx

Panning is a pretty huge deal IMO, and only having 3,5,7 or what ever points to use seems so restricting. Just my .02.

I think someone else mentioned it but I figured it is worth reiterating... saying there is only 3 distinguishable positions or 5 distinguishable positions, while being dependent on panning law, HAS much more to do with your speaker position than anything else.

If someone only feels like they can only really hear 3 positions, try moving your speakers apart more, AND check for reflections that might be skewing your stereo image. Those will unknowingly have a greater impact on your perceived stereo image than the pan law. And Pan law can be "compensated" for with speaker width. If you are using a -6 pan law and feel there is a "hole" in the middle, start moving your speakers closer together and you'll notice the hole goes away. If you are using -2.5 and everything seems to be very congested in your stereo field, spread the speakers apart and you'll start to notice MORE discrete panning locations across the stereo field.

[okydoky]

Originally Posted by Etch-A-Sketch

I'm not sure what you are trying to say here, I don't think you fully understand how the dithered mixer works... Any channel assigned to ANY output has 24bit dither added to it, regardless of whether it is an internal bus or an actual physical output. When anything is assigned to a bus or an output, a 48bit mixer gets created for it. That 48bit mixer outputs a 24bit signal regardless of whether it's internal or external. And when using the dithered mixer, that 24bit output has 24bit dither added to it.

Routing channels to an aux for panning is no different than routing channels to a physical output. When using the dithered mixer, all 48bit mixers get 24bit dither added to them on output.

From your last post it sounds like you are assuming that the audio within PT HD is 48bit. It isn't, only the mixer is. The TDMII architecture is only 24bits wide and the TDM chips can only process 24bit words. So, while the mixer is 48bit (56 with the 8bit accumulator), audio from disk, audio passed from plugin to plugin, audio outputted to a bus or hardware output, and audio received at the input of an audio track or aux (from either a hardware input or a bus) is ALWAYS 24bit.

hi,

i believe there are a few erroneous statements in your post about how the 48-bit mixer works, but i am just going to suggest that you go back and re-read the paper digidesign published and see if that is helpful

and rather than going over this numerous times with you, i will post once again the quote from the digidesign paper, and you can respond directly to that, rather than to me, in your next post [but please read it before you respond]. we are not talking about what goes on in pro tools [allegedly or factualy]. we are talking about what happens at the dac [interface hardware dac].


48-bit mixer, at page 6.

"We should take a moment to distinguish between the result of lowering a fader within the context of the internal mixer (where 48-bit precision is maintained) and lowering the level of a signal that is being sent to a DAC output.

As with any digital audio system (not just Pro Tools), if you
lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."




right.

[RKrizman]

Originally Posted by oky****

better panning and resolution is one of the advantages of mixing out of the box.

According to your analysis, better panning and resolution will occur if you stay in the box, since it is going out the DACs that causes the loss of resolution.

-R

[Etch-A-Sketch]

Originally Posted by oky****

i believe there are a few erroneous statements in your post about how the 48-bit mixer works,...i will post once again the quote from the digidesign paper, and you can respond directly to that, rather than to me, in your next post

48-bit mixer, at page 6.

"We should take a moment to distinguish between the result of lowering a fader within the context of the internal mixer (where 48-bit precision is maintained) and lowering the level of a signal that is being sent to a DAC output.

As with any digital audio system (not just Pro Tools), if you
lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."




right.

OK... you are taking page 6 out of context without reading pages 2 and 3.

more specifically, pg 3, 1st paragraph, "...Within the protools TDM mixing environment, when the dithered mixer is used, dither is added to the 48bit signal before it is reduced to 24bits to be placed back on the TDM bus and sent to the DACs. Dither is not added on a track by track basis, as this would produce unwanted accumulated noise. Dither is only added once at the master fader output of the mix bus so the total system dither noise is 3dB at -144 dB...."

Second paragraph, same page, "...the signal must be 24bit in order to pass from DSP to DSP via the TDM bus, but it's important to understand that the signal is around -144dB, which is below the noies floor of the converters. Consider that our ears can deal with a dynamic range of around 120dB (from threshold of audibility to the threshold of pain). This gives you an idea of how low "-144dB" is in terms of audibility-you'd have to have your monitor system cranked up to extremely high levels to even hear any sound that might exist in a noise floor this low. The same could be said about the lower 24bits. The sample values represented by the lower 24 bits in a 48bit words are between -144 and -288dB, so the dither only affects the signal at -144dB or lower."

And in the 7th paragraph of the same page, "it is important to understand that while some lower bits of the 48bit word are truncated when reducing individual channel levels, the quantization (or rounding error) that occurs when they are lost adds about one millionth of a dB to the total quantization error."

Pg 4, Paragraph 2, "In the pro tools mixer, input signals are multiplied by coefficients representing PAN AND GAIN as they enter the mixer, which produce 48bit words from teh original 24bit words..."

You keep telling us to "please read the white paper" but it really seems like you need to read it and not just skim it.

[unfadeable]

Im sure ill be tagged as an idiot for this one but please for the love of god someone answer this question... For a test i have a normal mono vocal track in PTLE 7.2. As i pan to the left and go slowly all the way over to <100 it sounds as if its still at '0' in the dead center! When panned to the right the sound doesn't sound at all like im panning to the right monitor at all! It just sounds like ive turned the volume down from 0.0 db to - whatever and is almost inaudible. There is nothing wrong with my NS-10's but im starting to freak out that i never noticed this before so i plugged a dynamic mic into an analog mixer>ns10s i and panned from left to right and it sounded like panning should.... I could go from left to right no problem and even at 100> the signal was 100% equal to the source ONLY comming out of the right monitor. So why is this happening to me inside the box??? What the hell is the solution??? This is crazy i never noticed this before errrr

[adpz]

Originally Posted by unfadeable

What the hell is the solution??? This is crazy i never noticed this before errrr

I don't exactly understand what you are describing here - but it sounds like a signal-flow and/or routing problem. You've got to break everything down and start again making it patch by patch.

[adpz]

Originally Posted by SeanG

As far as truncating one side or the other isn't it a moot point? How many times do you have all audio coming from one side and nothing on the other?

That's very much the point. Of course you don't have ALL the audio coming that way however, if you have sources panned to the left side you are going to have relatively low-level information on the right side, and if you have things panned to the right, you 'll have that low-level info on the left. This agglomeration of low-level information - to my ears - makes for a more congested sounding middle - or a more murky middle - than I hear if I do my pans OTB.

It's not my intention to have another OTB ITB battle royale. However, the panning difference is very clear in my setup - it's not subtle at all. It has nothing to do with pan laws or my monitors. If anyone sat in my chair here they'd hear the same things. The basic thing is that the panning in PT - at least in the stereo mixer - seems lumpy and indistinct and unfamiliar compared to the panning through the Neve summing box. If Digi or any other DAW could get the panning to sound as "normal" as that, we'll that would be a HUGE plus. That it doesn't yet work/sound like that is an additional obstacle that - at least for me - needs to be overcome when working ITB.

[okydoky]

Originally Posted by Etch-A-Sketch

OK... you are taking page 6 out of context without reading pages 2 and 3.

more specifically, pg 3, 1st paragraph, "...Within the protools TDM mixing environment, when the dithered mixer is used, dither is added to the 48bit signal before it is reduced to 24bits to be placed back on the TDM bus and sent to the DACs. Dither is not added on a track by track basis, as this would produce unwanted accumulated noise. Dither is only added once at the master fader output of the mix bus so the total system dither noise is 3dB at -144 dB...."

Second paragraph, same page, "...the signal must be 24bit in order to pass from DSP to DSP via the TDM bus, but it's important to understand that the signal is around -144dB, which is below the noies floor of the converters. Consider that our ears can deal with a dynamic range of around 120dB (from threshold of audibility to the threshold of pain). This gives you an idea of how low "-144dB" is in terms of audibility-you'd have to have your monitor system cranked up to extremely high levels to even hear any sound that might exist in a noise floor this low. The same could be said about the lower 24bits. The sample values represented by the lower 24 bits in a 48bit words are between -144 and -288dB, so the dither only affects the signal at -144dB or lower."

And in the 7th paragraph of the same page, "it is important to understand that while some lower bits of the 48bit word are truncated when reducing individual channel levels, the quantization (or rounding error) that occurs when they are lost adds about one millionth of a dB to the total quantization error."

Pg 4, Paragraph 2, "In the pro tools mixer, input signals are multiplied by coefficients representing PAN AND GAIN as they enter the mixer, which produce 48bit words from teh original 24bit words..."

You keep telling us to "please read the white paper" but it really seems like you need to read it and not just skim it.

hi,

no. you are still not understanding it, and i think you just don't want to. i have read it, believe me, and what i am saying is not news in some circles.

all of the quotes that you cited deal with signal before it hits the dac.

you are not understanding that the dac itself has to see high level in order to function at its potential.

so it does not matter how elegantly [or inelegantly] pro tools handles [or allegedly handles] the audio before it goes to the dac.

you need to speak directly to the quote that i cited and not try to debunk it by finding previous portions of the paper.

the reason the quote i cited is in the paper after the other information is to alert people to the fact that the level of audio in the dac itself is also significant, and the final determanitive factor.

the quote i cited from page 6 comes after the quotes you cited from pages 2 and 3, and it is already informed by the information you cited from the previous pages. the cite from page 6 already takes into account the information from pages 2 and 3, and it says, essentially, "all that said, you still cannot output low level audio from the dac without compromising the performance of the dac." essentially, "you have to send full level code to the dac if you want it to operate at its best signal to noise ratio" they say,

"if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."

put another way, you don't give a darn about what the audio is doing inside the computer, you care what it sounds like when it is converted back to analog for resampling.

you are trying to say: "i don't care what it says on page 6. page 6 must be a lie, because the information on pages 2 and 3 would mean something different if page 6 did not exist". what you are not seeing is that the information on pages 2 and 3 is only part of the story [up to the dac], and the information on page 6 is a final piece to the puzzle.


again [fifth time posting this quote]

"48-bit mixer, at page 6."

"We should take a moment to distinguish between the result of lowering a fader within the context of the internal mixer (where 48-bit precision is maintained) and lowering the level of a signal that is being sent to a DAC output.

As with any digital audio system (not just Pro Tools), if you
lower a fader assigned to an individual DAC output you can
lose resolution and dynamic range, as you are simply reducing the number of bits representing the output signal in comparison
to the DAC’s theoretical 24-bit resolution. You would be throwing away dynamic range that the DAC can represent
if you, say, pulled your fader down to -14 dB and left it there
so that peaks hitting the converter were far below full scale. That would decrease your signal-to-noise ratio significantly,
so optimizing signal levels going to individual DAC outputs is important to preserving signal quality—you want to get as much resolution out of the DAC that it can offer before clipping to take maximum advantage of its dynamic range performance."

if you want to respond, then respond to the quote directly, without trying to say that it does not mean what it says because of something you found earlier in the paper. i am not trying to tell you what to do, by the way, just trying to get you to the actual issue, rather than endless argument.

you also have to understand that digidesign is very adept at writing copy that minimizes any problems with their products, and they are generally unwilling to discuss that type of thing. so the fact that they have included the information on page 6 is significant.

i think you may be right about my idea with the auxes, and that it probably would not be any more effective. it was just an idea, but i should withdraw it since i do not really know that it would work.


right.

[okydoky]

Originally Posted by RKrizman

According to your analysis, better panning and resolution will occur if you stay in the box, since it is going out the DACs that causes the loss of resolution.

-R

hi,

no, that's what you are saying.

i am saying that better panning and resolution with occur if you mix out of the box and pan out of the box.

i am further saying that is obtuse to say "o.k., then you just have to stay entirely in the box at all time" to avoid the panning issue, which is what you are saying, essentially.

i am saying that it is obtuse because that is not the way the world works. people are mixing "out of the box" all the time, because it has numerous and formidable advantages, and it allows them to utilize enormous resources that are not available "in the box".

so, even if you can potentially partially eliminate the panning issue by staying "totally in the box" [and you do have to stay absolutely, totally in the box to do that], you rob yourself of the advantages of working "out of the box" in doing so .

i am saying that a proper solution to the problem is to craft a solution that includes not only the minority of people who choose to work "totally in the box" at all tiimes, but also to include the majority of people who wish to have the option of working out of the box as well.

i am saying that you need to specify that what you are advising includes the requirement that the audio be bounced directly to disk, and that no hardware inserts are being used, in order to ensure that you eliminate the "panning issue" to the greatest possible extent.

i am also saying that you may not be able to keep all of your audio channels above digidesign's claimed "-90dB point" using your method, and at that point you method would break down on you. that would not be an issue out of the box.

i am saying that your posts are a bad-spirited effort to put words in someone's mouth, in an attempt to justify your own desire to work "totally in the box", in order to validate your apparent need to feel that it is the best way to work, and to promote the idea that everyone should join you in your decicsion to work "totally in the box".

that is not going to happen, by the way.

as an aside, to the extent that some may cite analog noise floor as a concern, i am saying that it is by nature less obnoxious than digital distortion, and therefore preferable in that sense.

also, i am not recommending the use of crappy outboard gear or crappy consoles, but, rather, a good console and good outboard gear.

i am saying that "in the box" mixes can also be good and effective if you stay within the limitations that such a method has to offer, which a lot of people do, and i am not bothered by people doing that, or by doing that myself.

however, in the larger sense, trying to limit the entire music business to working "totally in the box" just to accomodate digidesign's panning issue [or just because "totally in the box is what you do in your home studio setup, or elsewhere] is immature and unrealistic.

so i am acknowledging that you may be able to get around the some of the "pro tools panning issues" by staying "totally in the box", and that may be reasonable for certain types of work.

but i am further offering a solution that also allows people to avoid the "pro tools panning issues" while working out of the box. in this way i hope to offer information and a solution that everyone can take utilize, not just people who want to restrict themselves to working "totally in the box" at all times.

but i don't care if you want to work "totally in the box" at all times. its your box. o.k.?

right.

[h4nc0]

oky****,

From reading your posts, is it fair to say that ITB setup with analog summing without level/pan control will suffer most from this problem, right?

[okydoky]

Originally Posted by h4nc0

oky****,

From reading your posts, is it fair to say that ITB setup with analog summing without level/pan control will suffer most from this problem, right?

hi,

i am not totally sure that i understand your question.

could you describe the analog summng setup a little [how many outputs being routed out, and to what?]. what do you mean by "ITB setup" and "analog summing without level/pan control?

i'm not trying to be funny, i am just trying to make sure i understand what you are asking, and i will answer as best i can.


right.

[h4nc0]

Originally Posted by oky****

hi,

i am not totally sure that i understand your question.

could you describe the analog summng setup a little [how many outputs being routed out, and to what?]. what do you mean by "ITB setup" and "analog summing without level/pan control?

i'm not trying to be funny, i am just trying to make sure i understand what you are asking, and i will answer as best i can.


right.

oky****,

It's ok, English is my second language so please pardon me on being unclear. I am talking about hybrid setup that are common these days. ITB systems (ie, protools or native) with analog summing (usually with 8 to 16 D/A) such as Dangerous, Folcrom, and etc that have no panning/level control and designed to just sum out of box. Does this make sense now?

[okydoky]

Originally Posted by h4nc0

oky****,

It's ok, English is my second language so please pardon me on being unclear. I am talking about hybrid setup that are common these days. ITB systems (ie, protools or native) with analog summing (usually with 8 to 16 D/A) such as Dangerous, Folcrom, and etc that have no panning/level control and designed to just sum out of box. Does this make sense now?

hi,

so you mean you are doing any and all level changes and panning in the box, and the analog summing is just a collection of passive analog channels routed to a single 2-buss with a volume control? there is no panning available before the 2-buss. correct?

can you route an individual channel to one side of the 2-buss or the other in any way?

does the 2-buss have a panpot? or ganged faders [is there individual control of the left and right channel levels at the 2-bus?].


right.

[h4nc0]

Yes. Let me just give you the links to those products.

Dangerous 2-BUS

Roll Music Systems

And many more, but these two are what comes to my mind atm.

[okydoky]

Originally Posted by h4nc0

Yes. Let me just give you the links to those products.

Dangerous 2-BUS

Roll Music Systems

And many more, but these two are what comes to my mind atm.

hi,

yes, using those products would not help the problem because you have no way to pan or adjust levels at all.


right.