Wav masters vs. DCP MXF Volume Discrepency

[ChungleDownBim]

Hey all - I'm a re-recording mixer with DCP curiosities

I've noticed a small and consistently measurable volume discrepancy between my masters (uncompressed interleaved WAV 24-bit 48khz) and MXF audio pulled out of my DCPs from DCP-o-matic and run back through Protools loudness metering (Youlean)

The discrepancy is:

DCP MXF = WAV Master -3.0 db

The evenness of that number made me think that there was something I don't know about in the workflow that's causing this.

Please flame me if it's something obvious. I'm investigating because I've had a client complain about a larger perceived discrepancy between the approved mix and their experience at a certain festival in Utah recently. I don't think -3.0 db is an explanation, but I'm determined to get to the bottom of it. Any comments on possible causes for larger (>6.0 db) discrepancies that can happen in the DCP side of the process also appreciated!!

<3 CDB

[carl]

Hi, that's strange - I can't think of a reason why they shouldn't the same, sample-wise. Can you send us a short example of WAVs and the corresponding MXF? Email to carl@dcpomatic.com if that's easier.

[Carsten]

Unless you do a 1:1 pass through in DCP-o-matic (as opposed to using the different ways of gain adjustments or up-mixers), there shouldn't be the slightest variation between uncompressed LPCM and DCP MXF.

I'm wondering wether some of the algorithms used for sample rate conversion or 16->24Bit interpolation MIGHT induce some gain changes, but they should be very very small.

But if you have 24Bit/48kHz in, DCP-o-matic shouldn't touch any sample values at all.

I think we had a similar question here a while ago, and I did a basic test that showed that at least with version I used, there wasn't any difference between source and output.


Now - a difference between mix and festival listening experience - that I would call the norm! The tools we have at hand are only good as far as minimising this difference, but not eliminating it. Even if calibrated properly, every venue will show it's own specific response to the same material. Common calibration techniques can only go so far.


I would also look at Youleans method of measuring across channels and time. Aside from loudness checking - you can also perform a classic evaluation of sample values between source and MXF audio. Loudness analysis is a complex algorithm and may show different results between source and DCP audio because of opaque conditions involved.

I'd say, compare your source WAV, and create a 24Bit/48KHz WAV from the MXF audio using simple tools (I use the free audacity), then look at both signals e.g. in Audacity, and you will probably notice that all samples are 100% identical. If the the MXF is analysed differently with Youlean - this may show some influence of the analysing methods, but nothing in the files.

[ChungleDownBim]

Hey there,

Thank you for the replies y'all!!

I've managed to account for this discrepancy - something in Adobe encoding world is taking everything down -3.0 db - when I import DCP MXF audio directly into Protools, the waveforms are 100% identical and measure the same on every loudness meter. It was my mistake for using Adobe as the intermediary - and perhaps this can serve as a warning to anybody who might use it as such, there's somthin in there that's lowering your mix slightly!

>Now - a difference between mix and festival listening experience - that I would call the norm! The tools we have at hand are only good as far as minimising this difference, but not eliminating it. Even if calibrated properly, every venue will show it's own specific response to the same material. Common calibration techniques can only go so far.

The bigger question, as to where my client's perceived discrepancy came from. I'm beginning to suspect a combination of buyers remorse on approved creative decisions & venue specific stuff. All that said, wondering if anyone has any experience with mix loudness disappointment here. I run my room low (76) - it's quite small. The mix was approved after screening in my room PLUS two others, one of which is newly renovated and really high end, screened at 82 - which is what the mixtech who runs that room recommended. All of this feels like it would have actually bumped nicely up to 85 when it made it to the festival...so maybe something else in their shorts program was wicked hot, and they turned the master fader down?

Another question (forgive my ignorance) - when shorts programs get put together, are festival tech directors re-encoding stuff to make one DCP for the whole program? Might there be room for error therein?

And, if the screening venue is setup for stereo, and they're delivered ONLY a 5.1 DCP (as was the case here), who/what is downmixing?

I'm wondering if bad downmixing might have come into play as well. The 2.0 webmix was also approved by director, but that was through my session downmixer, and was never delivered to festival. Director specifically complained about both the physical AND online screenings being too low - which leads me to believe somebody festival-side at some point downmixed us/the whole program....

Thanks for reading and for the input!

<3 CDB