Downmixers....
 

Got a little burnt recently down mixing 5.1 content to stereo. I’m exploring options in the downmix realm. Not sexy, but a reality.

Had some delay issues with a 5.0 avid downmixer...
and I love justin at cargo cult. He’s a genius. But the spanner downmixer consistently makes lower loudness when I downmix. (And I love spanner. It’s awesome.)

What’s everyone’s downmixer of choice? Considering the nugen as I enjoy using halo and vis lm. I met the folks at nugen at nab this past year. Very bright folks. Any experience with halo downmix?

I don’t have as much time as I like to demo various options. I was hoping to get a jump on things with your input.

How can Spanner “consistently makes lower loudness”?
It is what you set it. User responsibility —- there’s no magic to it. It’s a channel/level router/mixer with a great UI which helps with setting up automation.

Halo Downmix is an algorithmic approach. Spanner is routing made easier (as in you could set that all up with auxes but this does it as one insert.)

I hear ya! That’s what I thought too! Try it for yourself. I took a 5.1 print master ran it through spanner. L & R at unity. Center down 3. Ls and Rs down 3. Just like you should. Resulting stereo consistently 2-3 dB lower when I compared it to the 5.1 trough VisLm. I mean, please, tell me where I messed up. I’d prefer to be wrong in this case! If I have a chance post post a screenshot tomorrow.

What I bolded is wrong. That's the first thing I noticed.
When sending a discrete center into Spanner 5.1/2 plugin, it respects the equal panning law which means, L and R are already getting a -3dB center signal each. If you are additionally pulling down the center -3dB, then you have obviously overcompensated.

You can check it with a signal generator. This is also true for Pro Tools (well modern versions): a mono bussed to a stereo will be -3dB on each left and right. Not full signal on both left and right.

You also haven't mentioned what you are doing with the LFE. In addition, there's always the chance of phase issues on fold down which will change the summed signal levels due to channel content or mixing decisions. Most of these should not be significant enough to throw integrated loudness but depending on duration and content, it has the possiblity.

I'll leave you a link to Netflix's fold down instructions as a guide
https://partnerhelp.netflixstudios.c...tices-v1-1#1_2

I'll have to investigate how surround content is weighted for the ITU integrated loudness standards.

Weighted +1.5LU

Yeah you definitely should not be pulling the center channel fader on the plugin down 3dB (if that is what you are doing). It is already down three with the pan law. There are stereo presets that will give you a good start point. I use Spanner all the time and have never had any issues with lower loudness readings.

A simple test will confirm. Send -20dBFS tone on a mono track to your 5.1 mix bus and then through Spanner. Panned hard left it should read -20dBfs. Panned center it should read -23dBFS. If it reads lower then something is amiss.

Thanks all for the input. Though I fully understand pan law, I completely had a duh moment and didn't think that spanner would respect pan law. I mean, of course it would! Why wouldn't it!?! I guess I got confused as the avid down mixer needs a -3 center.... They day the mind is the first to go.


Anyway, I'm still looking at the NUGEN Halo down mix. Any experiences?

May I ask, what were your delay issues with Avid's Downmixer?

I use that one fairly regularly and cannot recall running into issues.

Just out of curiosity I did a test exporting the same pretty complex 5 minute segment of one of my last features using the Avid Downmixer, the Nugen Halo Downmixer and Spanner. They were all each set to their 'Stereo ITU' presets.

The loudness readings for all of them ended up being exactly identical, down to the decimal point (tested using both the Pro Loudness Analyzer and the Dolby Media Meter).

Then just for fun I did a null test between the three (inverting the phase). For most of the sections the Avid Downmixer and the Halo Downmixer phase cancelled entirely. When I tested either one against Spanner the signal was almost entirely gone, but I could very very faintly hear SOME signal. The one section of my export that didn't phase cancel exactly between any of them was one section that was extremely drenched in surround reverb, but that might be due to how the reverb plugins process and generate their signals (the reverb was live running through the plugin, not pre-printed)

In any case the loudness measurements between all three were identical, using their standard ITU downmix settings. So I'd choose whichever one has the best layout or interface for you, or based on how complex you want to tweak your parameters and settings etc. But they should all do the job they need to do just fine functionally.

Thanks for testing and sharing the results!!

Thanks for this. I just fell into the trap of assuming the trim was the only thing going on with the Nugen plug. The ITU preset is actually a net -6 in the centre channel and surrounds I believe.

hm, down-mixing is such a simple DSP process (summing signals and sending them to an output) I think ANY plugin (and even no plugin at all) can be used to do this. Just like an MS-XY encode/decode is always the exact came process and same result essentially a sum and difference.

I don't see why ANY of these plugins would render a different result with the same settings or sound different.

This.
If a LoRo downmixer behaves differently it’s broken. It is a really basic process.
The difference in reverbs would be that it’s a non-linear process so cannot cancel completely. Same would happen with dither, as that’s random noise no signals that have dithering applied would cancel to zero but around -60 to -80db dependent on the dither algorithm.

Net -3.
Not net -6.

Spanner, Avid Downmixer, Nugen all do the same.

Right, just saying it comes with an ITU preset that has the trim set at -3, on top of the internal -3 setting. In order to get -3 you need to make sure the trim is set at 0. Or at least that's my understanding from this.

Nope.
The ITU preset in Avid Downmixer shows center at -3 dB but the result is a -3dB per L and R.

The ITU preset of Spanner shows center at 0 dB change with a result of -3dB per L and R.

If you use the presets for both of these plugins and do not change anything, you will get the right result. It is always -3dB for Center as found in Left and Right, -3dB for Ls and Rs in the respective L and R channels, and LFE muted.

As I linked earlier, Netflix is fine with adding LFE in at a low level for their stereo fold down (so not the official ITU fold down)

Sorry for the confusion, I was just talking about the Nugen. Understood about the others.

agree that's because the AVID one is a down-MIXER with no pan pots where all signals are hard patched to the output path and just a fader for level (basically like hard routing each input channel in PT to a stereo out (with no panner) and then setting the levels according to the downmix parameters.
Spanner however is a PANNING-plugin with the earlier mentioned pan-law applied but an ADDITIONAL mixer in there as well.

Both render the same result when used properly just like the "naked" PT mixer would. The only difference is that Spanner is much more flexible and can do all formats unlike the rather limited AVID downmixer that doesn't even support all formats (7.1 to 5.1 for example)

The Avid downmixer, with me, is 0.7 dB hot in ITU 775
Confirmed over 300 episodes!!