Timecode on video shoots

[Roy Bemelmans]

Being a film buff I’m currently studying TC on the set. I know how TC is used on film shoots, but I was wondering how it is used on professional video shoots (whether they be SD or HD). So far I can't seem to get a clear overview.

What’s the TC situation..

- When recording audio with an audio recorder and using one video camera?
- When recording audio with an audio recorder and using multiple video cameras?

Is the audio recorder used as master or will there be a dedicated master TC generator? Are timecode slates used? Is TC recorded to all tapes? Do the cams just record TC onto one of their tapes' audio tracks or actually slave to incoming TC? What framerates are used? What about genlock?

Any help would be appreciated.

Thanks

[Ty Ford]

Lot's of ways.

If the camera outputs TC, you run a cable from that to your timecode recorder

Or, you jam sync either the camera to recorder (or vice versa) and don't leave them connected.

There's a LOT to know about timecode. There's a guy last name of wolfe who has a book on it. He hangs out in the r.a.m.p.s. newsgroup.

24, 25, 29.97ND, 29.97DF, 30ND, 30DF. You can lose yourself for hours (well, I did) trying to deal with all of the permutations.

Different poeple work differently. In a studio, the cameras all lock to a central house clock. In the field, I've heard of people using time of day timecode among multiple cameras and fixing slight diffferences on the timeline in post. That presumes the clock in each are set and are close to the same time. If they aren't you hope for a standard offset and apply that.

Regards,

Ty Ford

[Roy Bemelmans]

Thanks Ty.

Originally Posted by Ty Ford

Or, you jam sync either the camera to recorder (or vice versa) and don't leave them connected.

Am I right that if you jam sync the camera to the audio recorder, the audio recorder's TC wil actually determine the speed of the camera or just the internal TC of the camera? Will this TC be recorded by the camera on tape as LTC on an audio track or VITC or in some digital form? How will the NLE use this to marry audio and video?

Originally Posted by Ty Ford

24, 25, 29.97ND, 29.97DF, 30ND, 30DF. You can lose yourself for hours (well, I did) trying to deal with all of the permutations.

I know all about these, being an aspiring film composer.

Originally Posted by Ty Ford

Different poeple work differently. In a studio, the cameras all lock to a central house clock.

Locking (cables) or jam-syncing? What equipment is used?

Originally Posted by Ty Ford

In the field, I've heard of people using time of day timecode among multiple cameras and fixing slight diffferences on the timeline in post. That presumes the clock in each are set and are close to the same time. If they aren't you hope for a standard offset and apply that.

So you would just take a cable coming from the TC output of the audio recorder and take it along all cameras one by one to jam-sync to?

[Seth Bloombaum]

Roy, the answer to many of your questions is "it depends."

Ty outlined the basic techniques and procedures. One generator is the master, everything else either slaves to it or is jamsynched to it. The master can be a camera, an audio recorder, a slate, or a house generator.

However...

Much of what used to be standard procedures are crumbling. Few of the mini-DV generation of camcorders support TC in, out, or jam. I'm most familiar with Sony, e.g. the PD150, 170, FX1, Z1 - none of them support the TC functions detailed above. Sony wants you to step up in cost to do so.

AND, you need to step up in the cost of your playback deck as well.

Having said all that, TC is recorded as part of the digital stream on all these cameras, and their internal generators support preset, record run, and free run. And when you digitally transfer to common low-end and indie NLEs, that timecode is available.

Matching TC was EXTREMELY important when we were doing match-in editing, the first generation of computer-based editing in which a proprietary computer with proprietary software was controlling 2 or more players and 1 recorder. All this control was via external TC synchronizers, and decks were complicated and expensive so playback could be nudged to get them into sync.

For most people, it's a different world now. Incredible technology in the spare bedroom, etc.

No mistake - All timecode functionality is supported in BetaSP, DigiBeta, Film, pro audio recorders from Nagra to DAT to disk and other similar acquisition.

For people shooting their feature with a Z1 or DVX100 it's history, except as regards syncing dual-system audio. That's where time-of-day timecode is cheap and dirty, accurate to perhaps half a second if done with care, with tight sync to be done in the NLE. The better way is with the sound recordist's timecode slate - show the slate at the beginning of every take, that becomes the audio/visual sync point. But really in that use it's just a slightly more sophisticated clap board.

[Douglas Spotted Eagle]

Just to echo what Seth said....all the things we used to require in the past are slowly fading away, and in the future, they'll likely be so much less important, as there are other tools usable, and techniques being developed every few months.
"Bus Girl" had zero timecode on the set. it was all synced in post. That's not common, but likely to be moreso in the future.

[Steve House]

Jumping in, "jam sync" simply means the TC clocks in the two devices are set to the same value at the same time and then allowed to run independently. The camera and the audio recorder each have their own clocks. If we set them both to read 00:00:00;00 and start them simultaneously, assuming the clocks run at the same rate, they will stay in sync without a connection between them. If I'm recording, the timecode on the camera tape will match the timecode on the audio tape at any given time allowing me to match audio and video in post. Neither camera nor audio recorder controls the speed of the other after the clocks are set and started. Some TC systems allow the jam to be refreshed every so often so minor differences in run rate can be corrected but that's not universal. Some prosumer cameras like the Canon XL2 make jamming multiple cameras together quite simple - just set the cameras to use the same remote control channel, stop their clocks and zero them out or set them to the same value, arrange the cameras so they can all see the same remote control and then start the clocks all together with the remote. They'll all start off at once and on a well made camera like the Canon they should stay in sync over a reasonably long period of time before needing to be reset.

In film it works differently, at least in the classic systems - the camera outputs a synch signal that varies with camera speed. The audio recorder free runs at its own speed but records the camera sync into a separate control track alongside the audio. When the audio tape is resolved to perfed magnetic film in post, the recorded sync tone controls the speed of the dubbing recorder

Video cameras use the clock not only for recording the timing of the video through timecode but also to control the scanning on the image. If you have multiple cameras, the frames must be scanned in lock-step to each other so you can cut, dissolve, composite, etc the images from each without introducing glitches. (It isn't an issue editing from multiple source files in a digital NLE since the editing software takes care of syncing the frames). You can accomplish this by running all of them off of a common "house clock" through a process called Genlock instead of each camera using its own internal clock. This process also means that any timecode the cameras are generating would also run at exactly the same rate, sort of as a side benefit of each camera having its overall system clock slaved to the same house sync. This sort of arrangment usually involves the VTR's being separate from the cameras and also slaved to the same system clock. Thus timecode recorded to tape in such as system probably doesn't come from any of the cameras directly but rather is derived in the VTR from the same house clock that genlocks the cameras or from the house clock itself as would the TC recorded by any audio recorders in the system.

[Roy Bemelmans]

Thanks a lot for all your input, guys. Much obliged.

Clearly the situation varies from shoot to shoot..

Just to get a bit of an overview when working with say, multiple video cameras and a HD audio recorder:

Say you were to jam-sync your cameras to the LTC output from your HD audio recorder. AIUI, during jam-syncing the cameras would all convert this incoming LTC to a digital form that is recorded with the digital video stream on tape/disk (so the TC will no longer be LTC, no audio track on the video tape is used for TC, no VITC either). Then when importing this video via Firewire or SDI, the NLE will interpret this digital TC from the stream and create a timeline to it, syncing multiple video streams to this timeline? Timestamped BWF files from the HD audio recorder can then be dropped in and will snap to the right TC locations.

Is the above correctly put?

Also:

- How long a connection between devices is normally needed for proper jam-syncing?

- Do any cameras cross-jam? (e.g. from 29.97 input to 23.976)

- With 24p/23.976p shoots/editing and audio recorders being able to handle 23.976 TC, will 23.976 audio TC become more prevalent (instead of 29.97)? Anyone using 23.976 for both audio and video?

- What's more common in a studio using a central house clock, jam-syncing or actually continuously slaving to the house clock (cables)?

- AIUI, genlocking is basically only required when working with multiple digital video systems, i.e. mainly live broadcasting, right? So on e.g. a 24p sitcom it would't be necessary, right?

[Steve House]

Originally Posted by Roy Bemelmans

...

Just to get a bit of an overview when working with say, multiple video cameras and a HD audio recorder:

Say you were to jam-sync your cameras to the LTC output from your HD audio recorder. AIUI, during jam-syncing the cameras would all convert this incoming LTC to a digital form that is recorded with the digital video stream on tape/disk (so the TC will no longer be LTC, no audio track on the video tape is used for TC, no VITC either). Then when importing this video via Firewire or SDI, the NLE will interpret this digital TC from the stream and create a timeline to it, syncing multiple video streams to this timeline? Timestamped BWF files from the HD audio recorder can then be dropped in and will snap to the right TC locations.

Is the above correctly put?

Not quite ... Jam syncing sets the clock on the slave device to read the same as that of the master. But then the two clocks run independently of each other and rely on their individual accuracies to stay in sync after the instant of "jamming." In your scenario where the HD recorder is the master, all it does is tell the camera what time it is at the instant when the recordist jams it. From that point on, the two devices no longer talk to each other at all (until the next time the recordist jam's 'em), each of their clocks ticking merrily away on their own. Note that there are jam sync systems that monitor the slave's "health" and if there's a drop out etc it will automatically pick up the clock from the master to restore the sync.

Timecode does NOT cause any files dropped into the NLE to automatically snap to the timeline or to each other. It simply gives the editor a reference point on the audio and video tracks to help him line them up. It also is used when trimming the individual clips prior to dropping them into the timeline to mark in and out points or prepare an edit decision list. After all, if timecode caused the clip to jump to the matching time in the timeline, I couldn't intercut closeups shot today with a medium shot recorded last week nor could I intercut clips shot using free run timecode with clips shot using record run timecode.

- AIUI, genlocking is basically only required when working with multiple digital video systems, i.e. mainly live broadcasting, right?

Close - genlock is used whenever signals must be mixed or intercut in the analog realm. For example, the old-style editing required A and B roll source tape VTRs and a destination VTR. To dissolve from shot 1 to shot 2, 1 would be on the A roll and 2 would be on the B roll. You'd line up the two tapes and play them in sync through a video switcher on to the destination recorder. At the place for the dissolve, one signal would be faded out as the other was faded in. This required all three recorders (and the switcher for that matter) to be running at exactly the same speed with frame accuracy and all of the scan lines had to be happening at exactly the same time, hence genlock to lock them all together. Line 173 of the current frame on tape A had to be scanning at exactly the same time as line 173 on the current frame from tape B. Even straight assembly editing the source sync had to match anything already recorded on the destination tape in order to do the inserts without a jump in picture.

[Roy Bemelmans]

Ok, so let me rephrase a TC scenario involving multiple video cameras and a HD audio recorder:

Say you were to jam-sync your cameras to the LTC output from your HD audio recorder (free run). You take all the cameras by the audio recorder for jam-syncing one by one. During jam-syncing the cameras will all make their internal digital TC run in sync with the incoming LTC from the audio recorder. This digital form of TC is then recorded with the digital video stream on tape/disk (so the TC will no longer be LTC, no audio track on the video tape is used for TC, no VITC either). Let's say the HD audio recorder records to timestamped BWF files.

This is correcly put, right?

Then when compiling the dailies, you import all this video into your NLE via Firewire or SDI. What happens at this stage? How do you sync the different video streams and audio? I assume the NLE will interpret the digital TC in the video streams and create a free run timeline to this or something, syncing multiple video streams to this timeline? Timestamped BWF files from the HD audio recorder can then be dropped in at the right TC locations. You could then copy parts of this project to your final timeline / first cut..

Am I way off here?

Also:

- How long should a connection between devices normally last for proper jam-syncing?

- Do any cameras cross-jam? (e.g. from 29.97 input to 23.976)

- With 24p/23.976p shoots/editing and audio recorders being able to handle 23.976 TC, will 23.976 audio TC become more prevalent (instead of 29.97)? Anyone using 23.976 for both audio and video?

- What's more common in a studio using a central house clock, jam-syncing or actually continuously slaving to the house clock (cables)?

- AIUI, genlocking is basically only required when working with multiple digital video systems, i.e. mainly live broadcasting, right? So on e.g. a 24p sitcom it would't be necessary, right?

Thanks

[Roy Bemelmans]

Anyone? Eager to learn here..

[Steve House]

Originally Posted by Roy Bemelmans

...
Say you were to jam-sync your cameras to the LTC output from your HD audio recorder (free run). You take all the cameras by the audio recorder for jam-syncing one by one. During jam-syncing the cameras will all make their internal digital TC run in sync with the incoming LTC from the audio recorder. This digital form of TC is then recorded with the digital video stream on tape/disk (so the TC will no longer be LTC, no audio track on the video tape is used for TC, no VITC either). Let's say the HD audio recorder records to timestamped BWF files.

This is correcly put, right?

Almost. Assuming the camera records timecode, jaming will syncronize the clocks in the camera and audio recorder. During a shot the camera records its clock to tape as VITC between each frame. The HD recorder timestamps the BWF file with the starting timecode of the audio file from its own clock. LTC is more a feature of analog audio and DAT recorders AFAIK, not BWF, WAV, or MP3 files.

Then when compiling the dailies, you import all this video into your NLE via Firewire or SDI. What happens at this stage? ... I assume the NLE will interpret the digital TC in the video streams and create a free run timeline to this or something, syncing multiple video streams to this timeline?

This would depend on the NLE. In those I've familiar with, the editing timeline is generated by the NLE itself when you create a new project, not copied from video. When you add a BWF clip, it snaps into place to line up its starting timestamp to the NLE timeline.

[Roy Bemelmans]

Thanks Steve.

Originally Posted by Steve House

This would depend on the NLE. In those I've familiar with, the editing timeline is generated by the NLE itself when you create a new project, not copied from video. When you add a BWF clip, it snaps into place to line up its starting timestamp to the NLE timeline.

Suppose you recorded 24p to time of day / free run timecode. The audio recorder records at 29.97 fps and the camera at 23.976 (cross-jammed to eachother). What I don't get is the BWF files will correspond to 29.976 TC and need to be dropped in a 29.97 timeline, while the video has 23.976 TC. Can you work to 2 different timelines in NLEs (of course 29.97 and 23.976 will be the same to the second)? Also, your time of day TC is discontinuous and would be all over the place. So do you sync dailies on a take to take basis and then copy to a new project?

[Steve House]

Originally Posted by Roy Bemelmans

Thanks Steve.



Suppose you recorded 24p to time of day / free run timecode. The audio recorder records at 29.97 fps and the camera at 23.976 (cross-jammed to eachother). What I don't get is the BWF files will correspond to 29.976 TC and need to be dropped in a 29.97 timeline, while the video has 23.976 TC. Can you work to 2 different timelines in NLEs (of course 29.97 and 23.976 will be the same to the second)? Also, your time of day TC is discontinuous and would be all over the place. So do you sync dailies on a take to take basis and then copy to a new project?

Why would you do that? Set both the camera and the audio recorder to use the same TC format. Use the same format to create the timeline in NLE.

[Roy Bemelmans]

I thought most audio recorders didn't support 23.976 and a lot of audio post houses still work to 29.97..