slow performance and only 3 sound channels

[createmedia]

Hi

New to DOM 2.12.13 today.

Made a 2048 by 1080 XAVC.MXF Video + 8 channel audio file from a GV Edius timeline with 4096 by 2160 material on.

Bring file into DOM and set in DCP tab to 7.1 audio and put a diagonal row of green blocks to include all 8 channels. The graph is seeing audio info on all 8 channels.

Make DCP.

When I use the DOM player to play back the result I only get the first 3 tracks of audio, but it was set to 7.1. when I made the file.

To monitor the audio I have 7.1 creative labs audio set up on computer, which plays the channel ID wav file back correctly on a desktop player.

I am on a dual 16 core Xeon PC and I am surprised to find that the main app won't play the file back without stuttering and the player has to be at 1/4 resolution. Set to use all 64 cores. CPU underutilised during testing.

Am I doing something wrong here?

[Carsten]

The player may still have a 'bug' or incomplete mixdown scheme. I think there was some mentioning about including all channels in downmix, but I don't know wether that has been implemented already. I have played discrete 7.1 through HDMI audio on my Mac and a Windows machine, but that was a few versions back. Did you try to play back multichannel through your soundcard with the main app?

I see 'decent' playback from 8 cores up, so, don't know what's happening there. My dual 6core HT xeon machine get's close to full res without much stuttering. But, it's a software decoder, you shouldn't expect too much of it. Maybe the decoder has a bug with too many available cores.

- Carsten

[carl]

What operating system are you using?

[createmedia]

It is Windows 10 build 1803 with creative labs 7.1 card and 7.1 speaker system.

I have made an MXF DCI 2K Flat file with an embedded BWF audio file.

In the DOM Creator the file plays ok in real time with 2K Flat, it is a 4K DCI file it won't play in real time so that's the playback slow performance issue within DOM.

The audio graph shows all channels are there, but I can only hear content Channels 1,2 and 3 in the PC speakers. If I allocate content channels 3-8 to DCP 1 and 2 I still don't hear anything from those.

The Centre Channel comes out of both left and right speakers in DOM and the player.

The good news is that the transcode is going through at 23FPS almost real time.

The player in DOM is much better than the player app itself.

The player recognises the number of audio channels in the DCP but I only hear L,R & C and it drops frames even at 1/4 resolution. There is virtually no CPU activity during playback showing less than 2% .


Regards

Mike

[createmedia]

If I don't do any scaling the encode frame rate is 35FPS

[carl]

That's interesting... would you mind installing and trying this build to see if it's any better?

[Carsten]

As far as I know, multichannel audio interfaces can be setup as multiple separate stereo interfaces, or as a single 6 or 8 channel device. Depending on this occurrence, DCP-o-matic may only see two channels and downmix.

- Carsten

[createmedia]

Installed the 08F version which is just the same, I can't detect any difference.

The Wave Extensive test file, 8 channel identifier from Microsoft plays out of all the individual speakers in turn played from the desktop, so that is ok

The good news is that when I load the audio file (I chose 12 channel for 7.1) back into the Grass Valley Edius 9 Edit suite all the channels I allocated audio to it is all there so that is my only way of confidence monitoring.

Now I don't have a 12 channel audio card but I would expect to hear the audio that is present on the first 8 channels.

[carl]

Installed the 08F version which is just the same, I can't detect any difference.

Just to clarify: you are seeing no difference in playback or encoding speed with that version?

[Carsten]

DCP-o-matic never had issues packing all channels into the DCP/MXF - If you can see them in the audio graph (preferably under the DCP tab), they will be there.
The simple playback downmix function was never meant to be used for QC, it was to help in assembly of projects so you know where you are, hence the limitation to L/C/R.

We talked quite a bit about proper downmix methods here. In order to do it right, one would need a much more complicated audio setup (essentially, another playback audio matrix in setup prefs). Whenever I need to see what's in an audio MXF, I load it into Audacity and see the raw channels. But, as I said, DCP-o-matic never failed in creating the proper audio MXF.

What 16core CPUs/xeon exactly is your machine based on?


- Carsten