Scoring Central

We've talked a great deal all over the forum about stereo imaging, panning, and placement. One thing that seems to have escaped me somehow is that there are actually different kinds of panning at a basic level. (I'm sure we talked about it, but it probably just flew over my head, or I forgot). I just assumed we were all starting from the same basic place, but apparently not...

In switching my template over from FL Studio to Reaper, I've learned a number of important things. It turns out that FL's pan knobs (on the mixer and instrument channels) are stereo pan controls; that is, they shift the stereo image around the field. So, if you pan hard left, you get both stereo channels in the left speaker only, for example. Reaper, on the other hand, defaults to a stereo balance (what Sam was talking about in the link above), which just affects the volume of the side you're panning away from. FL also uses a circular panning law, which compensates for apparent gain boost and reduction when a panned signal passes from one side to center; Reaper, by default, does not.

I discovered this while trying to figure out why copying settings from my FL template to my new Reaper one resulted in such a wonky sounding mix.

Unlike in FL, however, Reaper allows you to change--even on a track by track basis--the kind of panning you want to you apply. Since it's what I'm used to, I went through and changed each track to stereo pan and used the included stereo balance plugin as an insert effect when desired. This is exactly what I had been doing in FL Studio, though I didn't quite know the particulars. I left the pan law unchanged (levels are relative anyway).

I've had issues in the past getting very roomy, very wide samples to sit well in a mix. Orchestral Brass Classic seems to always be present on the opposite side no matter how extreme I pan a patch. I pan the trombones 50% right and they are still all over the left channel. I've found that slapping a stereo balance (and even some side EQ) on these patches after initial panning really helps impart a sense of placement in a space. I generally don't touch the stereo width knob anymore for any reason.

I'm sure there will be some objections to this way of doing things, but here are my thoughts:

If a stereo balance pan pot just reduces the volume in one side, you are effectively cutting the patch in half the farther you get to the left or right extremes. You're losing stereo information that you might find necessary in, say, a violin ensemble. Affecting the volume of one of the channels after the stereo information has been shifted gives you better control when things are supposed to sound specifically closer to the right or left side of the space, but you don't want mono trombones.

I don't think there is a one-size-fits all approach here. It's nice to have options when working with a template comprised of libraries that have all been recorded in different ways and in different spaces. I've found this works for me, but I'd like to hear some other thoughts. I am fully aware that retaining stereo information in one side is kind of anti-stereo. I just figure that all I'm doing is trying to trick my ears anyway; whatever works is what works. It reacts with the reverb and moves to the opposite side anyway. (And I know we've gone on about this ad nauseam, I just feel like I've had a revelation!)

I've tried the Haas Effect approach. I've used the included plugin in Reaper to delay channels by a few milliseconds, but it's a bit over my head at the moment. And it sounds (quite literally) like it will have major effects on how I use reverb. It also seems to be more effective on mono signals than on stereo samples.

Video: Pan Modes (Reaper)

Thanks for bringing up this topic, it's definitely a very interesting one.

The one danger to mixing the left into the right and vice versa when panning is that you can end up creating comb filtering, if the mics are not in a coincident or near-coincident array. That's common with orchestral recordings which are usually a spaced pair or decca tree, both of which are non-coincident arrays and rely on timing differences. The result of this comb filtering can be pretty devastating in some cases, or almost benign in others. A lot of it boils down to distance, type of mics, position, and space.

You can think of two methods of stereo separation:
1. Phase (time) based.
2. Volume/timbre based.

Non-coincident arrays (spaced pair/AB, Decca, etc.) primarily rely on phase/time cues (like a stereo panning plugin using the Haas Effect), while coincident arrays (XY, Blumlein) primarily rely on volume/timbre differences (somewhat like most pan pots). Near-coincident arrays like ORTF, NOS, etc. use a mixture of both. Note that the off-axis coloration of microphones is never truly perfectly uniform (nor are two mics truly perfectly identical), so there will usually be slight timbre changes between the two channels. So, while all three types of stereo arrays create stereo signals, how exactly those signals "feel" is quite different (and often very subjective). In general, speakers/monitors seem to render near-coincident arrays less pleasingly than headphones do, at least in my experience and from what I've read from the observations of others. Coincident arrays too are theoretically impossible (the diaphragms of the microphones would have to physically intersect for a pure coincident array), with all real-life examples (at least in common use) being approximations.

Often I advocate mixing Haas Effect-type positioning and volume based panning (either method you discussed). In this regard, you treat the signal more like a near-coincident array, which I tend to find more lifelike in positioning. Just using volume treats it more like a coincident array, and just using Haas Effect treats it more like a non-coincident array.

Panning compensation is also a very interesting subject. A signal which is completely centered will seem louder than an equal signal which is only present in one of the channels. Thus often a -3 or -6 dB compensation is provided for center-panned mono signals. Indeed Reaper allows you to pick what kind of compensation you would like, even how it interpolates it I believe. Not that it probably makes much of a difference, but it can help perhaps.

(01-01-2021, 02:13 AM)Samulis Wrote: [ -> ]Thanks for bringing up this topic, it's definitely a very interesting one.

The one danger to mixing the left into the right and vice versa when panning is that you can end up creating comb filtering, if the mics are not in a coincident or near-coincident array. That's common with orchestral recordings which are usually a spaced pair or decca tree, both of which are non-coincident arrays and rely on timing differences. The result of this comb filtering can be pretty devastating in some cases, or almost benign in others. A lot of it boils down to distance, type of mics, position, and space.

You can think of two methods of stereo separation:
1. Phase (time) based.
2. Volume/timbre based.

Non-coincident arrays (spaced pair/AB, Decca, etc.) primarily rely on phase/time cues (like a stereo panning plugin using the Haas Effect), while coincident arrays (XY, Blumlein) primarily rely on volume/timbre differences (somewhat like most pan pots). Near-coincident arrays like ORTF, NOS, etc. use a mixture of both. Note that the off-axis coloration of microphones is never truly perfectly uniform (nor are two mics truly perfectly identical), so there will usually be slight timbre changes between the two channels. So, while all three types of stereo arrays create stereo signals, how exactly those signals "feel" is quite different (and often very subjective). In general, speakers/monitors seem to render near-coincident arrays less pleasingly than headphones do, at least in my experience and from what I've read from the observations of others. Coincident arrays too are theoretically impossible (the diaphragms of the microphones would have to physically intersect for a pure coincident array), with all real-life examples (at least in common use) being approximations.

Often I advocate mixing Haas Effect-type positioning and volume based panning (either method you discussed). In this regard, you treat the signal more like a near-coincident array, which I tend to find more lifelike in positioning. Just using volume treats it more like a coincident array, and just using Haas Effect treats it more like a non-coincident array.

Panning compensation is also a very interesting subject. A signal which is completely centered will seem louder than an equal signal which is only present in one of the channels. Thus often a -3 or -6 dB compensation is provided for center-panned mono signals. Indeed Reaper allows you to pick what kind of compensation you would like, even how it interpolates it I believe. Not that it probably makes much of a difference, but it can help perhaps.

Fascinating as always! I just don't know what to do these days. I listen to Poledouris' Conan soundtrack as a reference for a live orchestra, and pretty much look to Mattias or the Baldur's Gate soundtracks for VO references. Generally, I want to use the VO tracks as a goal to shoot for because that's what I'm working with. I don't have the kind of control over things that I would with a live recording. I kind of want to embrace VO's particular characteristics, but I find that my mixes always sound more condensed, less spacious, than the examples I listen to. I don't know if I want to chase after some unattainable ideal of "realistic," but I do want there to be some noticeable depth and positioning. I wonder how much of this has to do with the limitations of my setup.

As for destructive phasing and whatnot, I honestly never worry about it unless I hear something that's bothering me. I feel like phasing is one of those things people worry about precisely because they don't know how to detect it. How do I know it isn't there, ruining my mix? Better do all kinds of stuff I don't understand to keep it from happening! My rule is, if it doesn't sound like crap or completely disappear, it's probably not an issue. The thing is, how in the hell do you pan anything without it being somehow destructive to the original source? If the trombone section is recorded by a spaced pair with the listener directly in front of it, that's the only place it's ever going to sound "perfect"; but that isn't where trombones go in a mix.

Anyway, I wouldn't be so concerned about it if I had any musical ideas in the first place!

One has a way of coming to a conclusion relative to a given project...

Nayrb, it's funny you should bring this up because I've been doing some thinking along the same lines lately. Every since I started using Reaper I've used the Stereo Pan for stereo sources, and the default panner for mono sources. Most of the time it works just fine, but sometimes I feel the Stereo Pan has undesirable effects on the sound source. I don't know if this is simply down to pan law, but some instruments with a wide stereo image tend to sound sort of weaker and narrower the further from the center they are placed. This makes sense in a way, I suppose, since what is basically happening is that an entire stereo image gets shifted to the side. So that got me thinking. Are there any other ways to do it?

Turns out, there are.

ITDPanner is a panning plugin that uses a method called Interaural Time Difference (which is a form of delay-based sound placement I guess?) and I've been using this for the strings and brass in my new template. I don't know if it's just confirmation bias, but I actually think it makes a difference. Panned instruments sound more natural, and I'm not hearing that detrimental effect I get sometimes with Reaper's Stereo Pan. Again, I could just be imagining it, but so far I like what I'm hearing. It's a free plugin, so give it a shot if you're curious.

(01-03-2021, 03:58 AM)Mattias Westlund Wrote: [ -> ]Nayrb, it's funny you should bring this up because I've been doing some thinking along the same lines lately. Every since I started using Reaper I've used the Stereo Pan for stereo sources, and the default panner for mono sources. Most of the time it works just fine, but sometimes I feel the Stereo Pan has undesirable effects on the sound source. I don't know if this is simply down to pan law, but some instruments with a wide stereo image tend to sound sort of weaker and narrower the further from the center they are placed. This makes sense in a way, I suppose, since what is basically happening is that an entire stereo image gets shifted to the side. So that got me thinking. Are there any other ways to do it?

Turns out, there are.

ITDPanner is a panning plugin that uses a method called Interaural Time Difference (which is a form of delay-based sound placement I guess?) and I've been using this for the strings and brass in my new template. I don't know if it's just confirmation bias, but I actually think it makes a difference. Panned instruments sound more natural, and I'm not hearing that detrimental effect I get sometimes with Reaper's Stereo Pan. Again, I could just be imagining it, but so far I like what I'm hearing. It's a free plugin, so give it a shot if you're curious.

Thanks for the tip! I will check that plugin out.

Actually, I've had some pretty good results the last few days using time delay and stereo balance together. I'm using Reaper's stock Channel Time Delay plugin and the stereo balance mode on the mixer. Otto, who used to frequent the forum, is a proponent of time delay panning, and he advised me to start by plugging in a very small delay on the opposite channel (0.66 ms, to be exact). After that, I just adjusted the stereo balance by ear. I only did this for stuff on the far sides of the mix and stuff in the back that needed more audible space but also some kind of obvious sense of placement, too. I should also mention that I used the free TDR Nova plugin to slap a HPF on the sides of some of the wider, boomier percussion (my out-of-control True Strike timpani, for example).

All of this has been surprisingly effective. Like you, I was worried it would just sound good to me because it was different, but there is a definite sense of cohesion and none (that I can really hear) of the outright mangling of the patches that I was worried might occur. I get a sense of space and clarity that I just couldn't seem to get before.

I mixed on headphones (as I usually do these days). Like Sam says, there is a bit of a difference between the phones and the monitors. I haven't spent time "tuning" my room or anything, though, and I have closed-back headphones.

In any case, this example below is just noodling, and the panning is probably way too wide, but it's just kind of a proof of concept sort of thing.

Delay Panned Orchestra Noodling

(01-03-2021, 03:52 PM)Nayrb Wrote: [ -> ]In any case, this example below is just noodling, and the panning is probably way too wide, but it's just kind of a proof of concept sort of thing.

Delay Panned Orchestra Noodling

Wow, yeah! That does sound very big and roomy without being swamped in reverb. Yes, the panning is exaggerated leaving a big hole in the middle, but overdoing it when experimenting is usually the best way to get an idea of how something works. Very nice! When you say, "a very small delay on the opposite channel", do you mean the channel opposite of the direction you're panning the instrument? I.e. delay on the right channel of an instrument that goes to the left?

(01-04-2021, 06:15 PM)Mattias Westlund Wrote: [ -> ]

(01-03-2021, 03:52 PM)Nayrb Wrote: [ -> ]In any case, this example below is just noodling, and the panning is probably way too wide, but it's just kind of a proof of concept sort of thing.

Delay Panned Orchestra Noodling

Wow, yeah! That does sound very big and roomy without being swamped in reverb. Yes, the panning is exaggerated leaving a big hole in the middle, but overdoing it when experimenting is usually the best way to get an idea of how something works. Very nice! When you say, "a very small delay on the opposite channel", do you mean the channel opposite of the direction you're panning the instrument? I.e. delay on the right channel of an instrument that goes to the left?

I'm glad it translated to other ears! I feel like I've made a breakthrough in my applicable skills... Since I have to move away from my rig again (temporarily) in a couple days, I feel like I've achieved something that I can come back to and really get to work on new music with (finally). Yes, the middle is a hole, but what's cool about it is that it feels like I'll actually be able to fill it without making a cluttered mess of things. It seems like the space is finally there; it's the middle of somewhere. I reiterate that the stuff in the middle does not have a time delay on any channel; it's just the stereo balance. I think for smaller pans, that is probably all you need. I did add it to the perc in the back, though, which seems to add depth as well as positioning.

And yes, exactly as you said: the small delay would be in the channel you are panning away from. So there is a small delay in the right channel of the violins and a small delay in the left channel of the trombones. Then I pan with the stereo balance mode by ear. I'm at about 45% on the first violins here. I would say with a delay of 0.66 ms in the timing of one channel and a stereo balance pan at about 40% you are approaching something that sounds very exaggerated. I have no idea what other numbers to use on the time delay yet. It probably depends on the source, too. It seems that something below a millisecond is what you want, though. Otherwise it starts sounding like a delay or a maxed out ER on a reverb. The trick is to keep it below where our ears would really pick up on it. I can't say this with any certainty, but anecdotally, it seems like that 0.66 ms delay amounts to a perception of almost a 10% pan in one direction or another. I kept the number the same just to see how different panning percentages interacted with it.

Haven't had a chance to mess with ITDPanner, but I suspect something akin to the above is what's going on there. The description suggests it combines some kind of timing effect with a stereo balance. I'll see when I get a chance to try it out.

Cool, I'm going to give it a try!

OK, I've experimented with your method and compared it to ITDPanner. It's hard to do any scientifically valid A/B-ing since ITDPanner calculates the amount of delay automatically when you turn the balance knob, whereas with the Channel Time Delayer you need to enter the values manually and sort of guesstimate the right amount. But I would hazard that for all intents and purposes both methods are basically the same, and yields nearly identical results. I think I'm sticking with ITDPanner since it spares me the guesswork, but I definitely think you have a great sounding mix going on there, so I see no reason you should change what you're already doing

(01-04-2021, 09:37 PM).Mattias Westlund Wrote: [ -> ]OK, I've experimented with your method and compared it to ITDPanner. It's hard to do any scientifically valid A/B-ing since ITDPanner calculates the amount of delay automatically when you turn the balance knob, whereas with the Channel Time Delayer you need to enter the values manually and sort of guesstimate the right amount. But I would hazard that for all intents and purposes both methods are basically the same, and yields nearly identical results. I think I'm sticking with ITDPanner since it spares me the guesswork, but I definitely think you have a great sounding mix going on there, so I see no reason you should change what you're already doing

Cool! Well, that's good to know. I don't like to over-complicate things, either. If it works, it works. We can keep at it with our respective approaches and compare results.

Out of curiosity, do you slap that plugin on everything or just certain instruments? I've been trying to be deliberate and to choose wisely which instruments I make these changes to. But I'm not really sure yet of the best way to go about it.