How is SMPTE timecode recorded in digital video?

[A. J. deLange]

There is still a major distinction in that LTC is an audio bandwidth signal separate from the video signal and which, therefore, requires a separate channel for capture, processing etc. VITC is embedded in the video signal and so where it is employed no separate T/C connection is required. A secondary difference is that LTC persists troughout the frame and as it is a BPSK signal I'm guessing that the recovered carrier gives a continuous reference which can be used for rough sync (I'm pretty sure that systems which derive audio word clock from just LTC work this way).

[Steve House]

Originally Posted by A. J. deLange

Steve,

I'm with you on everything you say except that there is no relationship between samples and time code and between audio and video samples. One of the things that got talked about a lot in the past was that most of the prosumer cameras did not tie audio and video sampling together which potentially led to problems. In the XLH1 that has been fixed - audio sampling is locked to video (in the SD mode you have a choice of locked or unlocked though I have no idea what the implications of unlocked might be...

Understood - but are we talking about locked verus unlocked audio here? Does the frame rate setting of the timecode sent from an external clock or the camera to the audio recorder change the way the recorder's sample rate syncs to the word clock it derives from the incoming TC signal?

I don't want to put words in Roy's mouth but I got the impression he thought that setting the timecode to be recorded to audio or video to drop or non-drop, 30 or 29.97 or 24 or 23.976 fps would somehow affect the recording process itself and that when using double system sound, having the camera recording code at 23.976 fps but the audio recording it at 29.97 fps would make difficult or impossible to achieve or maintain sync. It's going to an inconvenience to find matching frames when the numbers on the video frame are showing, say, 01:23:45;12 while the numbers recorded with the audio matching that frame are 01:23:45;15 but hardly something to lose sleep over <grin>.

Roy gives the impression he thinks that when you drop a video clip with timecode into an NLE and then drop an audio file in alongside it that has the same timecode, they can automatically sync up, shifting and stretching or contracting the audio file so the the timecode numbers for each frame recorded by the audio recorder will align themselves to matching frame number recorded in the video - am I correct in reading your questions that way Roy? I'm not aware of any NLE's that do that but I admit I'm not familiar with Avid, FCP or amy number of others.

As an aside, the combo of a Canon XLH1 camera and a Sound Devices 744T recorder looks like it would be a dyn-o-mite package for shooting double system. And I imagine if you can derive TC from the LANC connector on an XL2 the two would be a good match as well. The 744's timecode module is by the Ambient folks, basically a Lockit box mounted in the recorder, and they claim an accuracy of <0.2ppm. It can serve as either a master or a slave. When syncing to external code but with the camera and recorder set to different frame rates the Ambient and SD folks say in their manual that the recorded code jams at each full second.

[Roy Bemelmans]

Originally Posted by Steve House

I don't want to put words in Roy's mouth but I got the impression he thought that setting the timecode to be recorded to audio or video to drop or non-drop, 30 or 29.97 or 24 or 23.976 fps would somehow affect the recording process itself and that when using double system sound, having the camera recording code at 23.976 fps but the audio recording it at 29.97 fps would make difficult or impossible to achieve or maintain sync. It's going to an inconvenience to find matching frames when the numbers on the video frame are showing, say, 01:23:45;12 while the numbers recorded with the audio matching that frame are 01:23:45;15 but hardly something to lose sleep over <grin>.

Oh no, that's not what I thought. Obviously everything can be synced. Hell, 23.976 video TC and 29.97 audio TC will be exactly the same to the second..

Originally Posted by Steve House

Roy gives the impression he thinks that when you drop a video clip with timecode into an NLE and then drop an audio file in alongside it that has the same timecode, they can automatically sync up, shifting and stretching or contracting the audio file so the the timecode numbers for each frame recorded by the audio recorder will align themselves to matching frame number recorded in the video - am I correct in reading your questions that way Roy? I'm not aware of any NLE's that do that but I admit I'm not familiar with Avid, FCP or amy number of others.

Well, I come from a music background. When I import timestamped BWF-files into my DAW (I use Cubase SX), I can automatically snap them to the right location when working to a timecode timeline. This method is used often in post to transfer audio tracks from one DAW to another or to an NLE.
Now say you use 23.976 TC for both your cameras and the audio recorder on the set. The audio recorder records to timestamped BWF-files (with embedded TC start position for each take). Fom what I've gathered so far the video location TC is shown by the NLE in a window.
Here's the idea: if the NLE can read the video TC, it could automatically move all the BWF-files I'm importing into it straight to their correct postion in relation to the video.. (making syncing your dailies very easy)

Originally Posted by Steve House

As an aside, the combo of a Canon XLH1 camera and a Sound Devices 744T recorder looks like it would be a dyn-o-mite package for shooting double system.

The 744T indeed looks like a very complete product, Steve..

[Steve House]

Originally Posted by Roy Bemelmans

...
Here's the idea: if the NLE can read the video TC, it could automatically move all the BWF-files I'm importing into it straight to their correct postion in relation to the video.. (making syncing your dailies very easy)
...

That it would! But I'm not aware of any NLEs that snap video to the timeline based on imbedded timecode the way BWFs snap to time. That doesn't mean they don't exist, but I'm not aware of any that do. I double checked on the features in Steinberg's Nuendo, which is relatively high end DAW software oriented to film and video sound work and didn't find any indication it would auto-sync video and audio like that. Maybe someone knowledgeable in Avid or Pro Tools will chime in on them.

[Roy Bemelmans]

Originally Posted by Steve House

That it would! But I'm not aware of any NLEs that snap video to the timeline based on imbedded timecode the way BWFs snap to time. That doesn't mean they don't exist, but I'm not aware of any that do. I double checked on the features in Steinberg's Nuendo, which is relatively high end DAW software oriented to film and video sound work and didn't find any indication it would auto-sync video and audio like that. Maybe someone knowledgeable in Avid or Pro Tools will chime in on them.

Just to be clear, I don't mean snapping the video to the timeline, but rather snapping the timestamped BWF-files to the video location TC (embedded in the video stream). Your timeline will just be continuous 'program TC', while the BFW-files snap to the Time Of Day TC embedded in the video dailies.. (provided the same TC was used for cameras en audio recording during the shoot)

Anyone know of NLEs that can do this?

[Steve House]

Originally Posted by Roy Bemelmans

Just to be clear, I don't mean snapping the video to the timeline, but rather snapping the timestamped BWF-files to the video location TC (embedded in the video stream). Your timeline will just be continuous 'program TC', while the BFW-files snap to the Time Of Day TC embedded in the video dailies.. (provided the same TC was used for cameras en audio recording during the shoot)

Anyone know of NLEs that can do this?

Another thought ... the BWF's timecode that is recorded in the file header is for the start of the file only. The code for the rest of the audio clip is not part of the BWF file. Now imagine this scenario - a 5 minute shot is recorded in dual system with camera and audio timecode sync'ed. Code on both audio and video recorders started at 00:00:00;00 and ended at 00:05:00;00. The video clip records the code for the entire clip. The audio file only records it begins at 00:00:00;00 - nothing else. In the NLE we first trim the video clip, cutting 1 minute at the head and the tail, and then add it to the timeline. The embedded code in the clip as it sits on the timeline now starts at 00:01:00;00 and ends at 00:04:00;00. We now add the audio BWF file. BUT the only TC reference in the BWF file is the fact it starts at 00:00:00;00, a code that no longer exists in the video clip at all. Where should the NLE put the audio clip?

[Roy Bemelmans]

Originally Posted by Steve House

Another thought ... the BWF's timecode that is recorded in the file header is for the start of the file only. The code for the rest of the audio clip is not part of the BWF file. Now imagine this scenario - a 5 minute shot is recorded in dual system with camera and audio timecode sync'ed. Code on both audio and video recorders started at 00:00:00;00 and ended at 00:05:00;00. The video clip records the code for the entire clip. The audio file only records it begins at 00:00:00;00 - nothing else. In the NLE we first trim the video clip, cutting 1 minute at the head and the tail, and then add it to the timeline. The embedded code in the clip as it sits on the timeline now starts at 00:01:00;00 and ends at 00:04:00;00. We now add the audio BWF file. BUT the only TC reference in the BWF file is the fact it starts at 00:00:00;00, a code that no longer exists in the video clip at all. Where should the NLE put the audio clip?

I see what you mean, Seve. But I guess you'd typically use the method we're talking about for preliminary syncing of dailies, then do your trimming..