View Bug Details

IDProjectCategoryView StatusLast Update
0002443DCP-o-maticBugspublic2023-12-31 03:45
Reportermhm Assigned Tocarl  
PrioritynormalSeverityminorReproducibilityhave not tried
Status resolvedResolutionfixed 
Target Version2.16.51 
Summary0002443: Always make 8 or 16 channels in mxf audio files
Description

Both Netflix (https://partnerhelp.netflixstudios.com/hc/en-us/articles/4417542010387-Digital-Cinema-Package-DCP-Specifications-Requirements ) and Deluxe ( https://hpaonline.com/wp-content/uploads/2022/07/Deluxe_Source-and-DCP_Delivery_Specifications_v5-11_20220314-2.pdf ) mandates to produce either 8 or 16 channel audio mxf files only (and pad with silence "black" audio).

[Deluxe]: "Audio MXFs shall always have number of tracks in multiples of (2). For combining with other pre-existing
elements and/or reels, Audio MXFs shall be padded with “black” to either (8) or (16) channels."

[Netflix]: "Audio MXF files must always have an even number of tracks (i.e. divisible by 2). Audio MXF must have either (8) or (16) channels, with unused channels containing silence/MOS."

Does DoM currently do this, or does it produce a six channel audio mxf for a 5.1 sound track?

Best,
Mattias

TagsNo tags attached.
Branch
Estimated weeks required
Estimated work requiredUndecided

Relationships

related to 0002393 closedcarl EasyDCP 4.1 says wild track format (channel config 4) must have 16 channels. 
related to 0002487 acknowledgedcarl DCPs made by DoM fail EasyDCP verification 

Activities

carl

2023-02-11 23:19

administrator   ~0005470

Last edited: 2023-02-11 23:20

DoM currently makes a 6-channel MXF for 5.1. I don't understand what Deluxe means by "for combining with other pre-existing elements and/or reels". Do you understand it?

This Netflix constraint also seems strange: "Audio MXF files must always have an even number of tracks (i.e. divisible by 2). Audio MXF must have either (8) or (16) channels, with unused channels containing silence/MOS."

What's the difference between a track and a channel here?

mhm

2023-02-12 02:17

reporter   ~0005471

I believe that HI/VI on channel 7/8 is more ore less mandatory in North America. I think (but I may be wrong) that Deluxe means that because of this you should always use 8 or 16 to be able to combine mxf files/reels that do not contain anything in channels 7/8. Also the same applies to be able to merge with mxf files not containing Dbox, sign language etc.. (eg distributor logos etc).

In the Netflix case I think track and channel are used to interchangeably to mean the same thing.

At least Deluxe got A LOT of field experience of what works and do not work. So even if we don't know the exact reasons it may be sound advice to always make 8 or 16 channel audio mxf i DoM and pad unused channels with silence.

Or what do you think?

Best,
Mattias

mhm

2023-02-12 03:28

reporter   ~0005472

Possibly also related to issue 0002393

When reading section A.1 of ST 429-2:2020 you could interpret WTF (Configuration 4) that it indeed requires all 16 channels to be present. Both Deluxe and Netflix seem to agree thoug that either 8 or 16 channels is ok. (Or maybe I am reding it wrong...)

Also Deluxe says in their document:


3.4.3 AUDIO FORMAT
Audio MXF shall utilize the “Configuration 4” Channel Assignment UL as noted in SMPTE ST 429-2:2020 [see https://doi.org/10.5594/SMPTE.ST429-2.2020], commonly referred to as “WTF” or “Wild Track Format”. No other channel formats, including “Undefined” or “0” (IOP), nor other formats described in SMTPE ST 429-2:2020, will be accepted

carl

2023-02-12 13:53

administrator   ~0005482

Last edited: 2023-02-12 13:54

I don't think any of this is really clear, to me at least. Perhaps the best we can do is see what passes "validation" by the latest Easy DCP version, since that's what everyone seems to be using for "QC" anyway. It would be easy enough to always use 8 or 16 channels, and I guess relative to the video file it's not a terrible waste of space.

carl

2023-02-12 13:56

administrator   ~0005483

It looks like either Deluxe copied Netflix's "guidelines", or the other way around, since they both have this weird statement about MXFs having an even number of tracks, and also having either 8 or 16 channels.

carl

2023-03-29 19:28

administrator   ~0005593

In SMPTE mode we now always write 16 channels since bb1c1b89260cf36c621f7f2b471eb23f2ff15b0c

mhm

2023-06-21 05:12

reporter   ~0005768

Maybe change these to only make 16 channels when there is audible information on any of channels 9-16, and otherwise make 8 channels? I thinkt it is a bit superflous to always make 16 channels.

Also, this is how most of the big labs do it in my experience.

carl

2023-07-25 10:05

administrator   ~0005857

https://dcpomatic.com/forum/viewtopic.php?p=10715

carl

2023-07-25 10:09

administrator   ~0005858

The thread linked above gives a pretty good summary of the arguments against always writing 16 channels

  • it seems to (probably) be a Deluxe/Netflix workflow thing rather than any standard (though the relevant standards are not clear to me)
  • it's wasteful (16-channel DCP is 17% larger than a 6-channel with 100Mbit/s video)
  • lots of commercial SMPTE DCPs are stil being mastered with 6 channels or whatever (e.g. the trailer for OPPENHEIMER)

mhm

2023-07-25 11:49

reporter   ~0005859

Maybe the compromise I mentioned above is a way forward? I.e. either 8 or 16 channels. But only 16 channels when some of the channels 9-16 are actually used.

Indeed all 7.1 DCPs I have seen have 16 channels (with unused padded with silence).

The number of channels I have seen in DCPs in the Wild:

2: Stereo, not OK. But mostly works
6: 5.1-mixes, okayish.
8: 5.1 or 5.1 w/ HI/VI
16: Everything else (7.1, Atmos, Dbox, ...)

Eliminating the first to, and use 8 or 16 "dynamically" is a good compromise in my opinion.

carl

2023-07-25 11:59

administrator   ~0005860

It might be the best way. If we did this I guess the DCP channels drop-down would be used to select the number of actual MXF channels. I think this is still useful, if only to decide how many channels are offered for mapping in the audio matrices.

carl

2023-07-25 12:00

administrator   ~0005861

@carl branch channels-again respects the user's choice.

overlookmotel

2023-10-14 12:56

developer   ~0006008

To clarify, am I right in thinking the policy is now:

  1. Allow any number of audio channels, as specified by the user (2, 4, 6, 8, whatever).
  2. Warn the user before they make the DCP that 8 or 16 channels is recommended, if they chose some other option?

From observation, I think that's what's happening in DOM 2.16.65. But just wanted to check I have this right.

overlookmotel

2023-10-14 13:03

developer   ~0006009

@mhm One question: You say above "2: Stereo, not OK. But mostly works".

I'm interested by the word "mostly". Do you have any in-the-field experience where a 2 channel DCP failed to play back correctly?

I think we saw that problem once, but it was hard to disentangle whether the 2 audio channels was the root cause, as there may have been other problems with the DCP or the venue's projection equipment.

I ask because it's not uncommon to receive 2-channel DCPs (Resolve produces them, for example if the source is stereo) and am wondering if they should be rejected / rewrapped.

carl

2023-10-14 22:25

administrator   ~0006012

@overlookmotel your observation about 2.16.65 is correct.

overlookmotel

2023-10-15 20:08

developer   ~0006019

Thanks @carl.

mhm

2023-12-31 03:45

reporter   ~0006171

@overlookmotel Sorry for the late reply.

No, I have actually not seen any problems with this in the field. I mostly handle 2-channel DCPs at festivals and am also very careful to update all equipment to the latest firmware.

But I am always a bit nervous when receiving 2-channel DCPs because then the risk of something else being wrong is definately higher..

Bug History

Date Modified Username Field Change
2023-02-11 23:09 mhm New Bug
2023-02-11 23:19 carl Assigned To => carl
2023-02-11 23:19 carl Status new => feedback
2023-02-11 23:19 carl Note Added: 0005470
2023-02-11 23:20 carl Note Edited: 0005470
2023-02-12 02:17 mhm Note Added: 0005471
2023-02-12 02:17 mhm Status feedback => assigned
2023-02-12 03:28 mhm Note Added: 0005472
2023-02-12 13:53 carl Note Added: 0005482
2023-02-12 13:54 carl Note Edited: 0005482
2023-02-12 13:54 carl Relationship added related to 0002393
2023-02-12 13:56 carl Note Added: 0005483
2023-02-25 23:39 carl Target Version => 2.16.45
2023-02-25 23:39 carl Estimated work required => Undecided
2023-03-03 22:51 carl Target Version 2.16.45 => 2.16.46
2023-03-05 20:36 carl Target Version 2.16.46 => 2.16.47
2023-03-08 00:27 carl Target Version 2.16.47 => 2.16.48
2023-03-15 15:45 carl Relationship added related to 0002487
2023-03-23 20:24 carl Target Version 2.16.48 => 2.16.49
2023-03-27 18:36 carl Target Version 2.16.49 => 2.16.51
2023-03-29 19:28 carl Status assigned => resolved
2023-03-29 19:28 carl Resolution open => fixed
2023-03-29 19:28 carl Note Added: 0005593
2023-06-21 05:12 mhm Note Added: 0005768
2023-07-25 10:05 carl Note Added: 0005857
2023-07-25 10:09 carl Note Added: 0005858
2023-07-25 11:49 mhm Note Added: 0005859
2023-07-25 11:59 carl Note Added: 0005860
2023-07-25 12:00 carl Note Added: 0005861
2023-10-14 12:56 overlookmotel Note Added: 0006008
2023-10-14 13:03 overlookmotel Note Added: 0006009
2023-10-14 22:25 carl Note Added: 0006012
2023-10-15 20:08 overlookmotel Note Added: 0006019
2023-12-31 03:45 mhm Note Added: 0006171