View Bug Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0001281||DCP-o-matic||[All Projects] Bugs||public||2018-04-22 01:02||2018-06-11 14:00|
|Platform||Mac||OS||OS X||OS Version||10.11|
|Target Version||2.12.x||Fixed in Version|
|Summary||0001281: No full audio downmix for multichannel audio|
Ooops - I was just checking multichannel DCP and audio file playback in DCP-o-matic 2.12.x and 2.13.x - the default playback device on my Mac is stereo.
Only input tracks assigned to DCP channels 1-3 (L/R/C) are audible during playback. When I assign input 4,5,6 (LFE, Ls, Rs) to either channel 1,2 or 3, they become audible. They are silent when assigned to DCP channels 4,5,6 (LFE, Ls, Rs).
Same happens for both DCPs, audio MXFs, and e.g. quicktime files.
Strange thing - when I try 2.11.17, only channels mapped to 1/2(L/R) become audible. In 2.12.x/2.13.x, channels become audible when mapped to either 1,2 or 3... Looks as if the downmix function is 'truncated' after channel 3/center.
I guess that is not intentional? ;-)
I may have been so fascinated about the option to play back to actual multichannel interfaces, that I didn't bother testing the downmix option - at least as long as the important center/dialog channel was always there... Probably the reason no one else complained so far. Have to test in windows.
|Tags||No tags attached.|
|Estimated work required||Unknown|
It was intentional, but probably wrong. Ls/Rs and Lfe are included in 79a9da3ed267f2fa95aa9615dd662a1cea2e9d81 (2.13.29)
How do you sum/normalize with multiple channels? Based on active matrix assignments? Because I guess we often have the scenario with 5.1 target channels, but only L/R and Center being populated. Downmix-Attentuating based on target channel numbers then would result in a very low level. But I guess we do not always have a dependable sample peak per channel to base downmix-attentuation on?
No I'm just doing
Lt = L (down 3dB) + Ls (down 3dB) + C (down 6dB) + Lfe (down 10dB)
It's not very sophisticated... and I'm not sure how much it matters. We certainly don't have enough sample peak information to do it exactly.
The nerd in me wants a separate audio matrix for playback only, separate from the DCP-oriented matrix. But I understand how much that will confuse users. It could matter at some point when people want to monitor playback from the player or DCP-o-matic main, because mixdown will alter FS reference. No big issue for now, but I had a discussion about this with a film maker on a german forum who wanted to setup his own monitoring/review room and have proper reference levels. I guess the only important thing for now in that respect is that, when playing out to multichannel capable audio devices, that DCP FS=channel FS. Most people operating on that skill level will understand that reference levels and downmix don't go together well.
|2018-04-22 01:02||Carsten||New Bug|
|2018-04-22 01:02||Carsten||Status||new => assigned|
|2018-04-22 01:02||Carsten||Assigned To||=> carl|
|2018-04-22 01:04||Carsten||Description Updated||View Revisions|
|2018-06-09 23:35||carl||Status||assigned => resolved|
|2018-06-09 23:35||carl||Resolution||open => fixed|
|2018-06-09 23:35||carl||Note Added: 0002473|
|2018-06-10 10:49||Carsten||Note Added: 0002474|
|2018-06-11 09:49||carl||Note Added: 0002477|
|2018-06-11 14:00||Carsten||Note Added: 0002478|