Mid-side decoder removes stereo separation?

Anything and everything to do with DCP-o-matic.
carl
Site Admin
Posts: 2362
Joined: Thu Nov 14, 2013 2:53 pm

Re: Mid-side decoder removes stereo separation?

Post by carl »

The current "mid side decoder" does

centre = (left_in + right_in) / 2
left_out = left_in - mid = (left_in - right_in) / 2
right_out = right_in - mid = (right_in - left_in) / 2
jimsky7
Posts: 5
Joined: Thu Feb 29, 2024 6:14 pm

Re: Mid-side decoder removes stereo separation?

Post by jimsky7 »

Thank you for that explanation. I erroneously thought that L/R were exchanged (i.e. L becomes R and R becomes L). The removal of the opposite channel (divided by 2) would result in the listener hearing 'greater separation' wouldn't it? In many cases that may be fine. When I mix sound for my films, I'm pretty precise about placement and in general my L and R channels are what I want the viewer to hear. You might ask why I don't do a 5.1 mix - and that is a good question.
carl
Site Admin
Posts: 2362
Joined: Thu Nov 14, 2013 2:53 pm

Re: Mid-side decoder removes stereo separation?

Post by carl »

I suspect it's all rather dependent on the source material and how far apart the speakers are from each other, but I don't know for sure.
Carsten
Posts: 2665
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: Mid-side decoder removes stereo separation?

Post by Carsten »

I'll do some tests.

- Carsten
Kewl
Posts: 84
Joined: Mon Jan 17, 2022 5:13 pm
Location: Montreal, Canada

Re: Mid-side decoder removes stereo separation?

Post by Kewl »

jimsky7 wrote: Sun Mar 03, 2024 7:17 pm When I mix sound for my films, I'm pretty precise about placement and in general my L and R channels are what I want the viewer to hear. You might ask why I don't do a 5.1 mix - and that is a good question.
Would you consider mixing in 3.0 (LRC)? Rather then fiddling with a MS processor in a stereo to 3.0 scenario (where the MS processing will always introduce polarity reversal signal in the resulting L and R channels), mixing directly in 3.0 would help in preserving your mixing intent.

It is much easier, and with predictable results, to downmix 5.1 or 3.0 to stereo, than trying to upmix stereo to 3.0. MS processing is a linear (passive) process that was originally designed for two-channel audio and using it in 3.0 is suboptimal.

Some parametric (active) processes are designed for upmixing and will give much better result in isolating center information of a stereo signal to the center channel without introducing polarity reversal signal in the L and R channels.
Last edited by Kewl on Tue Mar 05, 2024 9:04 pm, edited 1 time in total.
Kewl
Posts: 84
Joined: Mon Jan 17, 2022 5:13 pm
Location: Montreal, Canada

Re: Mid-side decoder removes stereo separation?

Post by Kewl »

Here are some other linear processes for stereo to three-channel conversion: http://elias.altervista.org/html/3_speaker_matrix.html
neilren
Posts: 3
Joined: Thu Feb 29, 2024 3:47 pm

Re: Mid-side decoder removes stereo separation?

Post by neilren »

Thanks everyone for your help on here, and especially to Carl for adding the 'create centre channel' option.

In case it's useful to anyone, here's what happened with my DCP...

After noting the apparent stereo separation issue with the mid-side decoder, and not wanting to risk the new 'create centre channel' option as time was short, I left my sound on the left and right channels only. Following the software's tips I made the DCP with 8 channels.

I delivered it to the cinema and went in the next day to watch/listen. The picture was perfect but the sound was inexplicably right heavy. I listened from various points around the auditorium and the front left and front right speakers only were working, as expected, but for some reason the right was much louder. From the righthand side of the cinema it sounded like a mono mix coming from the right speaker. Only from the lefthand side of the cinema did it sound like stereo. We tried it in another of the cinema's screens with the same result. Extensive analysis of the DCP's MXF audio file by me and the re-recording mixer found no explanation.

Fortunately the projectionist was kind enough to come in late the next day (the last day before my screening) to ingest a new DCP which I hastily produced. She told me that the cinema didn't have 7.1 so this time I made the DCP with 6 channels (5.1) and in case that wasn't the source of the problem I also used the mid-side decoder this time. I figured even if the left and right channels were unbalanced again the centre channel would help correct it a bit.

I'm pleased to say this DCP sounded exactly as it should, correctly left/right balanced and with stereo separation, and of course a centre channel.

So the moral of my story is: trust the mid-side decoder, but do check how many channels your cinema supports.
carl
Site Admin
Posts: 2362
Joined: Thu Nov 14, 2013 2:53 pm

Re: Mid-side decoder removes stereo separation?

Post by carl »

It's a bit of a mystery (to me, at least) why a 7.1 DCP with silence everywhere except on L/R should not play as expected on a 5.1 system.
Carsten
Posts: 2665
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: Mid-side decoder removes stereo separation?

Post by Carsten »

This cinema has an issue with their speakers. If your L/R levels are correct in DCP-o-matic's audio level graph, they the DCP is correct they probably have a technical issue. These issues are not rare, but are usually masked when playing mainstream 5.1 or 7.1 content.


You could play a test DCP there to pinpoint the issue.


- Carsten