Genlock for 2-cameras doing long takes - hardcabled or Sync box

[tourtelot]

Yes, the Grace m108 native clock is 10PPM.

 

Lynx Aurora (legacy) is 22PPM, Aurora (N) is 7PPM.

 

D.

[Matthew Steel]

On 8/9/2023 at 1:36 AM, tourtelot said:

Is this a real number?  Can you cite a device(s) spec'd at 0.1PPM?  Sincere question.  I find what are called "high precision digital clocks" spec'd at 1PPM.  There is a device that is touted as specifically being a "Dante Leader Clock" and is rated at 1PPM for clock accuracy.  One part per million over the length of one year.

 

No affiliation: https://studio-tech.com/wp-content/uploads/2022/04/ds_m5401a_5.pdf

 

And what clock accuracy is REALLY necessary for any work that you or I might do.  I have never seen any program that doesn't cut away to some other shot over the long length of time that has been mentioned.  At least in my world, picture gets cut to sound so a PPM among friends shouldn't matter.  Please correct me if my thinking is off.

 

Oh and Philip's thoughts about joining two Dante networks up for a multi-camera concert shoot?  Me neither!  I don't want the hassle of a pissing contest about who controls the joint network.  It was bad enough in the day arguing about who took the transformer side of an analog microphone split.  I'll use my analog splitter and take the transformer side.  You power the mics.  Or the other way around.  I really don't care.

 

D.

 

I pulled the 0.1 spec from memory - the device I was thinking of was a 744T which when I actually looked it up is "< 0.2 ppm when tuned with an Ambient Master Controller."  The master clock in my system is an original Digidesign Sync I/O.  Its factory calibration spec is +/-5ppm but can be calibrated to within 1/3ppm.  I have calibrated it as close as I can to my video group's 744T, which is probably not calibrated to anything particular but they trust it as accurate.

 

I do relatively little video but when I do, drift can be significant.  Some of the projects really do include long shots of live musical theatre where the only edit is for intermission.  These single shots are typically a wide shot of the stage and/or an inset of a conductor camera.  How drift is handled varies depending on the group dong picture:

* One group gets a reference mix from me and cut picture to it.  So they will adjust sync if needed, but less drift is better.

* The other group attempts to keep everything in sync from the start.  While they haven't exactly "complained" about drift, they have reported the varispeed ratio they used to align my audio to picture.  It is for their sake that I have tried to get things as tight as I can.

 

 

[tourtelot]

2 hours ago, Matthew Steel said:

Some of the projects really do include long shots of live musical theatre where the only edit is for intermission. 

 

 

But no one anywhere stays on the wide master for one half of a concert, right.

 

I get the wish to keep sync as tight as possible but the difference in "drift" between an audio rig (let's say a Dante network) and a multi-camera set up should be pretty minimal if clocks used are in the 1PPM accuracy range.  Even 5PPM.  Think about that. Sync accuracy within 5 parts per million. Over a 24-hour day?  Unnoticeable. Weird but I never hear complaints from editors in my world (live acoustic music performance streams) about sound/camera drift.  Maybe they just fix it, but in my experience, if post editors are inconvenienced in any way, they will let the field folks know.

 

Also to be clear, this is about "drift" and not about matching latencies.  I typically dial in audio delay to match video "lip sync" but once that is set, there are no sync issues that I have ever heard of.  Just my experience of course.

 

There are devices that will function as the master clock to a Dante network using black burst or tri-level sync and then set to "sync from external" to all the Dante cards in the network.  Dante will never get out of sync with itself; one of the parks of a Dante network.

 

But I am not sure I believe the expense is warranted in most cases.  A fifteen camera Rolling Stones concert shoot might be an exception.  But in that case, budgets would allow for negotiations.

 

D.

[Philip Perkins]

Actually we posties generally don't let the production people know about sync issues much, esp in concert work, we just fix them if we can and move on.  We don't have time for discussion about what's already done unless there is a huge disaster in progress.  These jobs are usually one-off, not a series where you can build a relationship, exchange test files and dial things in over a few tries.   We usually proceed from the point where everyone assumes that what's delivered to post is in perfect sync from beginning to end, and then fix what needs fixing (since "perfect sync" in a live performance video often doesn't LOOK perfect even if it is numerically correct TC-wise!).   There are always many places to pull sync up in a live show, but assuming that puts me in the uncomfortable position of adding work to someone else's gig without their permission.  Thanks for the info that Dante clock is consistent within the network--the issue seems to be getting the network in sync with a TC generator for the cameras to begin with.

 

[tourtelot]

9 minutes ago, Philip Perkins said:

the issue seems to be getting the network in sync with a TC generator for the cameras to begin with.

 

Fair enough, I hear you.

 

FWIW, all this seems to have very little to do with TC sync, since it seems TC sync is not nearly accurate as needed.  But more to do with video sync, gen lock which as I understand, can be delivered via black burst (which I am familiar with) or tri-level (which I know almost nothing about how that works).  I think, in the real world, clocking the Dante network to gen lock would be the closest to being ideal (expensive box) or, if I understand, cutting video to audio.  I assume that this is/can be done the other way around but when would that be necessary, and wouldn't it cause many more sync issues?

 

D.

[Philip Perkins]

I haven't been on Really Big Music Multicam shows lately, the ones I'm doing rely totally on TC sync with cameras (cameras deriving sync from ext TC with no genlock used), so the accuracy of the TC they get vs the audio recording is important, as is the audio WC being locked to the TC.  That's why I usually want to be the show masterclock, since my rig can make very accurate WC and TC that are locked together.  This works well for most shoots I do.  For bigger stuff the conversation with the FOH people about possibly supplementing their rig with an external box that has a super accurate clock, can make WC, TC and Dante can be delicate if they don't have this together already.

[tourtelot]

Yep.  The bigger the show, the more points of failure.  Hopefully, professionals all around will make it all good, but that isn't always the case.

 

D.