Conformed original BWAVs are out of sync by up to a frame compared to the AAF

[pindrop]

Or Avid starts dealing with audio at it's native resolution, samples. Avid has said this would be difficult for them to implement.

Yes again I was interested to know the mechanics of how Avid needs to deal with it, I know it needs to deal with audio in samples but doesn't there need to be a way of reconciling that, to the discrete frames of the picture? Even if Avid does deal with sound in samples, the picture remains in frames of say 24fps so a little piece of sound if it starts sooner than the frame boundary, or a frame of picture if sound starts after the frame boundary needs to be lost to reconcile the two, samples since midnight (a very fine division of time) and frames of picture (a much coarser division of time).

[Wandering Ear]

Patrick May tell us differently... it can't be that difficult.

You're right, i forgot, it's easy to do, and detrimental until done.

[Jeff Wexler]

It seems strange to argue that achieving a solution to this issue is not possible as Sound Devices have demonstrated that it is possible.

 

If I'm reading Take's post correctly, and I think I am, the conclusion is that even Sound Devices is not able to get it right every time, the math is just not there. So, to say it is not possible does not denigrate the efforts by Sound Devices to deal with this "problem" but it still is not the ultimate solution.

 

Please re-read Take's post:

 

I've been thinking about it for a while now. Simply stated this is a bug in AVID that cannot be completely solved by an audio recorder. We can do our best to get it correct in most cases, but it is mathematically impossible to always start a recording at a frame boundary.

 

It is pretty simple for the standard 24,25,30 fps and recording at 48000Hz to start the audio file at a timestamp that is a multiple of 48000 samples. Starting the audio file at a timestamp that is a multiple of the sample rate, which would start a recording at exactly the start of a second.

 

However if you would put this 48000 Hz file on a 29.98 fps time line 'actual seconds' do not align with frames boundaries anymore.

 

The problem is that a BWF file has no notion of the frame rate of the movie.

The second problem is that the timestamp is recorded as number of samples since midnight, so it is not possible to calculate the start of a frame (or second) without knowing the frame rate of the movie.

 

In the newest versions of Boom Recorder you can already specify the sample rates and frame rates how the file is recorded on the set and how the file will be used in post production. It will be important to set those correctly (it didn't use to matter a lot, because the post-production-frame-rate is not recorded in a BWF file) for the record-on-frame-boundary feature to work. Still, for drop-frame-rates I can probably never get it to work.

[Constantin]

Yes but picture frames are not in samples so doesn't there need to be a way to reconcile that with sound that is? So sound can start ahead of a picture frame boundary or after a picture frame boundary, so would Avid need to top and tail clips to the nearest picture frame boundary for which sound exists without moving the sound, and therefore clipping fragments (a frame of picture or some samples of sound whichever is the least) off both picture or sound to achieve this when placed on the timeline?

yes, that's what I thought. Again, I would say that others can do it too

[Constantin]

If I'm reading Take's post correctly, and I think I am, the conclusion is that even Sound Devices is not able to get it right every time, the math is just not there. So, to say it is not possible does not denigrate the efforts by Sound Devices to deal with this "problem" but it still is not the ultimate solution.

.

Yes, thanks for re-posting it, Jeff. I did read Takev's post the first time. But as far as I know, on most recorders TC and sample rate are not in sync anyway, so I'm not sure why their relation show that it isn't (always) possible. But, I may be wrong.

[Marc Wielage]

Yes again I was interested to know the mechanics of how Avid needs to deal with it, I know it needs to deal with audio in samples but doesn't there need to be a way of reconciling that, to the discrete frames of the picture? Even if Avid does deal with sound in samples, the picture remains in frames of say 24fps so a little piece of sound if it starts sooner than the frame boundary, or a frame of picture if sound starts after the frame boundary needs to be lost to reconcile the two, samples since midnight (a very fine division of time) and frames of picture (a much coarser division of time).

 

Remember also that the camera shutter has an effect on how we perceive motion. I have trained over a dozen dailies people for years and warned them not to consider the slate clapped until the sticks no longer "blur" on the screen. Having said that, the usual preference if we have to go forward or back is to be slightly early than slightly late. (At least, that's the way it used to be in the CRT days.) I always argued that slightly late looked better, but I was generally overruled by the editor. 

 

In digital, the camera shutter can be referenced to external sync (genlock), and so at least then there's a definite relationship on how each frame is being captured and the leading edge of sync and the specific frame of timecode. It "theoretically" should not drift with external genlock and timecode, assuming it's running at the right speed.

 

I think anybody who expects audio timecode and video timecode to be 100% perfect 100% of the time has unrealistic expectations. Even under the best of circumstances, I think it's only right about 75% of the time, but all shots still have to be checked. You check, you move on if it's right. If it's wrong, you slide it a frame, re-check, and move on. It's trivial and not time-consuming. Anybody who expects more than this has no concept of how post works. 

 

What I would worry about is if sometimes the sync is 1 frame out, sometimes it's 2 frames out, sometimes it's 4 frames out, and sometimes it's dead-on. That would drive me crazy. But I rarely (if ever) see that in a dailies situation. When I do, I quietly call the sound department, see if we can work out a fix, and move on. We don't call the post sup unless things are really falling apart (like a drop/non-drop situation). 

[patrickveigel]

.

Partick Veigel says:

" It cannot be that difficult. "

You're right that this is an assumption. But it is based on two facts:

 

1. as Constantin keeps pointing out Sound Devices have been managing to avoid this problem for years.

 

2. we never hand out the originally recorded files with Zaxcom recorders. They are always mirrored, i.e. rendered files. For this rendering you can specify all kind of parameters as TC pull up or pull down, the  file type, the amount of tracks of the file, the bit depth and a lot of other helpful changes. My assumption is, that it should not be too difficult to alter the length of the files in that stage so they start and stop on full frames or even on full seconds. Still it is only an assumption. 

 

I agree with everybody else who is saying that this should be fixed by avid but I doubt that this will happen soon. 

 

 

I think I understand Takev explanations but again I have no idea how SD have been managing with drop-frames.

 

Quoted from Matt Mayer from further above:

"Very early on we decided it was a good idea to start files on the 00 frame, for cross-jam and frame rate conversion reasons.  Also, in studying the most common playback equipment of the time (primarily the Fostex DV-40), all of their files started on frame boundaries, so it made sense to conform to that paradigm"

 

Now if this is the solution or it simply went unnoticed that SD files are also out of sync by up to one frame in drop-frame workflows, I don't know.

Living and working in Europe I've never recorded anything in a DF rate and I doubt, that I ever will.

 

So a fix in the mirror stage of Zaxcom recorders allowing full second files for non DF frame rates would totally do the trick for many if not most of us.

[Wandering Ear]

Patrick, Howy has said this is not a simple quick fix for him, the pre record software interferes and would have to be rewritten. what sounds simple often isn't.

Again, this is not an issue with Zaxcom, most recorders operate this way, this is an issue with Avid. It isn't an issue with the other major nle's.

I would also argue that most of the shifting that occurs is small enough to keep the sound/picture sync within an acceptable level.

[Constantin]

Patrick, Howy has said this is not a simple quick fix for him, the pre record software interferes and would have to be rewritten. what sounds simple often isn't.

Just for my own understanding as I am not a Zaxcom user: if the fix were implemented at the mirroring stage, as Patrick suggested, how could pre-record be an issue?

All the pre-recording would have been done already. Maybe, just maybe, Howy didn't think of implementing the fix at the mirroring stage?

[Wandering Ear]

Just for my own understanding as I am not a Zaxcom user: if the fix were implemented at the mirroring stage, as Patrick suggested, how could pre-record be an issue?

All the pre-recording would have been done already. Maybe, just maybe, Howy didn't think of implementing the fix at the mirroring stage?

You're right, the prerecord wouldn't affect the mirror process. However i think this is the wrong place to implement a patch (not a solution). Patrick may not, but i use zaxcpnvert on my primary cards all the time. As well as with my wireless.

In addition, i expect my mirror files to be exactly the same as my primary files. Truncating them and changing the ssm stamp is not the same. I can only begin to imagine issues if files were transfered some from primary, and some from a mirror card with truncation. Different start times and durations for the same file name (supposed to be the same recording)? sounds like a great way to create more problems by fixing something that isn't broken.

[macrecorder]

Well, it could easily be a checkable option in both mirroring and zaxconvert. However, it is a real issue in some edit suites, which is how this whole debate started, so I don't see how people can blithely assert that it isn't a potential problem. Maybe it isn't in most, but it is a workflow issue which deserves some attention, from whatever quarter.

[Jeff Wexler]

Okay, a workflow that deserves attention, I'll give you that, but this whole discussion seems to be going no where. One recorder, Sound Devices, has solved this "problem" and none of the others have, not Zaxcom, Aaton, Fostex, Sonosax, Avid, Boomrecorder, etc., etc.

So, how have all the productions survived. For me personally, hundreds and hundreds of sound rolls somehow miraculously are in sync with no complaints. Worthy of attention? Sure. But MY attention, I don't think so. And I'm not at all worried that I will lose any work because of some disgruntled overworked person in post begging me to use a Sound Devices recorder.

[macrecorder]

Neither am I, as it happens. That's not my concern.

[pindrop]

Okay, a workflow that deserves attention, I'll give you that, but this whole discussion seems to be going no where. One recorder, Sound Devices, has solved this "problem" and none of the others have, not Zaxcom, Aaton, Fostex, Sonosax, Avid, Boomrecorder, etc., etc.

So, how have all the productions survived. For me personally, hundreds and hundreds of sound rolls somehow miraculously are in sync with no complaints. Worthy of attention? Sure. But MY attention, I don't think so. And I'm not at all worried that I will lose any work because of some disgruntled overworked person in post begging me to use a Sound Devices recorder.

 

No matter the state of mind of that disgruntled overworked person in post and who or what they beg, they have nevertheless got to be paid and there has to be the work for them and if that work is generated by an issue that is machine specific, that might just get noticed and mentioned as an additional cost etc. etc. etc.

In addition the cause has now been identified, posted in a widely used forum, and become common knowledge whereas it was not anything like as clear before, so what might follow from this?

That even if the issue can be to some extent ignored, in a mixed manufacturer machine situation for example it is likely to get mentioned and might easily be inferred that some machines are inferior, and don't deliver as reliable material, not a confidence building scenario, and not one as we've discovered easily explained.

Bottom line; one machine delivers files without a problem the other machine does not, now choose!

[old school]

If I ever have trouble sleeping I will re read this thread. What a snooze. So much about nothing. My eye lids are getting heavy from the worry of one frame. ZZZZzzzzzz

CrewC

[pindrop]

If I ever have trouble sleeping I will re read this thread. What a snooze. So much about nothing. My eye lids are getting heavy from the worry of one frame. ZZZZzzzzzz

CrewC

More sarcasm to add to the pile, you could always re-learn the trick of editing what you read, you must have forgotten it somewhere along the line, you really don't have to read anything that doesn't interest you. Plus it's somewhat obtuse to declare it's all about one frame when it's clearly a wider issue but there you go perhaps this is really about something else. Don't we all ignore dozens of threads of no personal interest, so why the need to comment on this one not being of interest?

There does appear to be a wide variety of motivations on this otherwise wonderful forum, as in any playground, grandstanding, point scoring, one-upmanship, self-aggrandisement, self promotion, personal alliances, brand promotion, OCD alleviation, to name a few, take your pick? But very sorry if you have been bored.

[John Blankenship]

I just looked out the window and contrary to the unrelenting obsession of some, I'm happy to report, the sky is not falling... repeat, the sky is not falling. 

 

You can all go back to your regular pursuits.

[pindrop]

I just looked out the window and contrary to the unrelenting obsession of some, I'm happy to report, the sky is not falling... repeat, the sky is not falling. 

 

You can all go back to your regular pursuits.

You must be using the best sky-props but I've heard there's ones with problems that.........Oh no!

[old school]

Personally, I save my energy for what I can do, not what is out of my hands. Would I like a standard for all work flows and machines whether they are digital or analog? No brainer. But this has never been the case ever since sound was wed to picture. I choose to live with this reality. As for commenting on a topic, I... ZZzzzz..........

CrewC

[John Blankenship]

LOTD!

[Wandering Ear]

A bird just flew by me 50 feet away, and i just realized it's chirp is more than a frame behind what i see! The world must be imploding!

[Marc Wielage]

I think this topic has produced so much sarcasm and dismissal from so many because it just isn't a serious problem. There is just way too much real world experience with use of ALL the rest of the recorders that have this "problem" to take it seriously. I sincerely doubt that anyone has lost any confidence in the performance of their chosen recorder even with this new found knowledge of the "problem".

 

And we also shouldn't ignore the fact that the cameras are inconsistent as well. I guarantee you, if you stack up a high-end Sony 4K camera, a Red 4K camera, and an Alexa, feed them all the same timecode, then start and stop them all day... they'll be out of sync with each other in inconsistent ways. Crap happens. Post's job is to fix it. It ain't that hard. 

[thope]

Why can't sound mixers just learn to press the record button when the frames read exactly 00! How hard could it be? Might make the slow days more fun

[pindrop]

Why can't sound mixers just learn to press the record button when the frames read exactly 00! How hard could it be? Might make the slow days more fun

No really the best fix of all is to bury your head in sand, as recommended by experts, I'm off to try it.......

[patrickveigel]

"And we also shouldn't ignore the fact that the cameras are inconsistent as well. I guarantee you, if you stack up a high-end Sony 4K camera, a Red 4K camera, and an Alexa, feed them all the same timecode, then start and stop them all day... they'll be out of sync with each other in inconsistent ways. Crap happens. Post's job is to fix it. It ain't that hard."

 

You're right that crap happens and that it is the job of the assistant editor or the telecine operator to move the files in sync when loading into the Avid. That's what they're  doing. The project is edited and in sync. 

This thread has been started though because when the project is handed over to audio post and exported for the use in protools everything is moved out of sync again because of the frame based nature of Avid. Meaning the work of the assistant editor has to be done again by the dialogue editor. It has always been done like that one could argue but it doesn't have to. Just re-read the very first post in this thread. 

I'd rather have the dialogue editor to have more time to polish my tracks than to spend two days to do the work that has been done already.

 

"There is just way too much real world experience with use of ALL the rest of the recorders that have this "problem" to take it seriously."

 

At least here in Germany Sound Devices, Aaton Cantar and Zaxcom recorders are by far the mostly used recorders on film sets. Of these three Zaxcom is the only brand that doesn't start on frame boundaries.