XDCAM HD, Native 23.976, Multicam

Hello,
Is there anyone out there cutting double system 23.976 multicam?
I am on a project shooting two or three XDCAM HD (25Mbit, 23.976) and two 744T 4trk audio recorders. The spec is for a 23.976 HDCAM SR master, so we're offlining in FCP at 23.976 and sending EDL to the online house and OMF to audio.
At least that is the plan.
The problem we're running into are numerous. Some are technical and others are workflow or operator generated as none of the people involved are used to working in this framerate.
Mainly I find FCP's multicam problematic for audio. In editing we would like to be able to dump all the audio across (2x 4ch and 3x Camera audio) for every multicam shot and just switch the camera shots. So far that seems impossible. The Avid editor in me tried to just build synced sequences, but editing sequence to sequence is not something you can get fast at.

Another issue is that we have two methods for getting footage off the XDCAM HD. We can digitize a DV signal normally over firewire with a standard 2:3 pulldown added by the deck. Removing pulldown using Cinema Tools gets you a 24fps clip which will play in a 23.976 sequence. The other method is to copy the files off the disk as native 23.976 XDCAM HD files and what we're trying here is recompressing them to DV for less load on processors and memory. This second method is faster and more reliable than the first and avoids alot of confusion.
Because there is a problem with the TC of the audio systems we have to sync by waveform in a sequence, match frame, mark in, make an independant clip, and then multiclip by in point. It's really hit and miss though. I doubted the assistant was doing it right untill I did some of the clips myseslf and even though the sync sequence played fine the multiclip was way out of sync.

It's a fast moving improvisational type of show so the on-set crew are limited in how much they can change their process. We have a ton of other issues that we're dealling with but I would be interested to hear if anyone out there is doing anything remotely like this in FCP.

Thanks.
andrew

We have 2 Dual 2.5GHz G5s with 4.5Gig RAM for 2 editors and 2 assistants.



Post Edited (07-29-06 10:02)

ak
Sleeplings, AWAKE!

> The Avid editor in me tried to just build synced sequences, but editing
> sequence to sequence is not something you can get fast at.

"Sequence to sequence"? Why not just put the angles as individual clips, synced up in one sequence? You only have two or three angles, right?

>"Sequence to sequence"? Why not just put the angles as individual clips, synced up in one sequence? You only have two or three angles, right?

It wouldn't be practical when you have 20-30 scenes per episode and long takes and 3 angles and at most 14 tracks of audio. I know I could probably get used to doing it that way but it's a last resort.
We have cameras and audio starting and stopping, inconsistent slating, all sorts of stuff. I am just looking for any workflow suggestions from people who have done this. We have to work in a pretty high pressure "producer over the shoulder" and "notes from the network" environment so some suggestions I have read on this forum about putting the camera angles in boxes by hand and stacking clips probably won't be practical.

thanks
ak

You mentioned that you had a problem with the TC of your audio machines. What specifically is wrong there? Was it a one time problem, or is it something you have to live with regarding the audio decks you used? If your TC can not be trusted with your audio decks, then I'd say that there needs to be some back up into consideration in production.

You also said that there was inconsistant slating and audio start and stops. Are you telling me that this production precludes the shooters and audio people from consistantly slating and/or recording the entire time?

From what I gather, it looks to me like you were handed a dog's dinner and now it's your problem. Multicam works great on FCP, but it requires some forethought in production or else you're going to have problems in post.

I do have one more question. You said that you were unable to lay all the audio into the timeline. Is that because of the start and stops of the audio mixer? If so, my suggestion would be to try and sync all the audio up first. Use whatever tracks you have that were recording the entire time as a base, and then sync up the broken audio by hand. Tedious I know, but like I said, they did this to you in production by not anticipating what you need in post. Once the audio is in place, you can multiclip all your cameras, and go, go, go!

Andy

I won't get into all the specifics of what's actually going on, discretion being the better part of valour and all that, not to mention public forums not being the place for that either.

That said, as we are shooting a thirteen episode season and have three in the can, I am trying very hard to work out a number of the problems before the next episode goes before the cameras. Ep 1 is, as you say, to put it nicely, a dogs breakfast as they started shooting two days before my start date, Ep 2 is less so as we're using the file transfer method of transferring the files of the XDCAM disks and have worked out some basic workflow issues with the time that bought us.

I thought that if we put all the devices at 23.976NDF and ran one of the audio units as master TC into the smart slate and Clockits on the second audio unit and the three cameras we'd be good to go. The cameras are in sync with each other and the slate, the audio units are in synch with each other but not the cameras. Waveforms line up fine and don't drift but the TC is always off, necessitating a laborious hand synching procedure. The TC problem is such that even with a hardwire TC connection between Camera and 744T the imported TC of the audio file doesn't sync. It's either a problem with the BWF conversion utility or something more inscrutable in the chain that I can't fathom.

In multiclip editing, if I have 5 angles, (3x video and 2x audio) how do I get all those tracks to go across in a basic overwrite or insert edit? By default I get 4 tracks matching the chosen angle which isn't enough. The cameras have scratch mixdowns from each four channel audio unit but alot of the time that isn't enough to build an offline mix for approval. This would be one workflow issue I would love to have explained to me.
Is there a way to make the two 4ch clips into one 8ch clip before multiclipping?

One thing I have learnt on this project is that tapeless production has many benefits (speed, picture quality) but the discipline enforced on assistants by the log and capture procedure, while slower, does have it's benefits. In the old method you never have disks full of computer-ese file names. Some of that could be fixed with a software polish on the part of Sony to allow you to rename the title field right on the XDCAM disks before you import them but the problem is really evident when you get 8gigs of compact flash cards full of sequentially numbered 4 track audio files.



Has anyone out there ever worked entirely in a 23.976TC workflow, shoot/capture/edit/online?
I know that some film and most video eng/efp audio people tend toward 30 or 29.97 and we might go to that if we have to.

thanks
andrew "dancing as fast as I can" kines



ak
Sleeplings, AWAKE!

There's a merge clip function that might help you. It allows you to combine 1 video track and up to 24 audio tracks. It's meant for workflows where audio is recorded on sources separate to the video source.

Could you not then combine all your audio sources to one angle? 3 camera angles with 4 trks audio each plus 4 trks audio each from 2 separate audio units would equal 20 audio tracks total (if I'm understanding you correctly). Connect them to one of your 3 camera angles and multiclip it to the other 2 angles. Lay the clip with 20 audio trks into the multiclip sequence and lock the audio tracks so they remain unchanged. Then can't you just switch and cut between video angles?

I feel like I might still be missing something about your situation because that seems overly simple. I'm not at my system so I can't really test what I'm typing, but it feels like it should work.

Andy

The cameras can record 4 tracks per clip but we have them set to 2. One is the camera mic which, hopefully, we should never have to use other than to confirm synch and the other is the rough mixdown from the 4tracks sent wirelessly by a "Modulus". A audio unit to the A camera, B audio to B camera.
So we have 3x2 =6 camera tracks and 2x4=8 audio unit tracks.

I'll look up that merge clip function when I get there next. That seems like the way to go. Without trying it out I would think that merging the 2 x 4-tracks with the A angle and then having only three angles in the multiclip is the method I'll try first.

>I feel like I might still be missing something about your situation because that seems overly simple.

You're right, it should be simple but dealing with a new format and new-to-me audio recorders, and a totally tapeless post system has made murphy's law very real. Little things going wrong in every area can make for a total mess.
We've even discovered a whole day of audio where one of the operators recorded the wrong speed (24 or 30 with no pulldown, I think) so nothing stays in sync for more than 10 seconds. That was almost the last straw.

thanks again
ak

It sounds very frustrating. Why is it that it takes a post-production person to notice that pre-production testing would be helpful?

Sounds like someone blew the wad on the new toys, but didn't consider that you might have to work differently to get it done correctly.

I don't know if you knew about this, but Apple and Sony have been going around the country touring and going over the workflow you're using. It's their XDCam HD and FCS tour.

Unfortunately, most of the tour dates are finished, but the remaining ones are here:

[seminars.apple.com]

I went and got a LOT of my questions answered. I don't know if you're anywhere near those tour dates, but if you were, I would HIGHLY recommend going and just grilling the Apple and Sony people.

Andy

Frustrating indeed!
Can't comment on all the issues but the biggest one of all has been solved for us as of this morning. The utility BWF2XML was doing inncaurate conversions of 23.976 TC for import into FCP. Andreas Kiel, who wrote the software has been a big help on tracking down the issue and has sent us a beta which addresses this issue. Big props to him for all his help.
I am testing it now but it seems to be working fine. Using "find" to gather all clips with similar start TC and shoot date, merge the two audio units together with the master angle by timecode and multiclip the three clips by timecode.
I have one multiclip question though. When I lay one into a sequence the result is a clip with out of sync flags on the audio tracks and the audio is out of sync. If you slip them everything lines up fine. Each 4track audio unit is out by a different amount. The amount seems to be related to the difference between the start of each 4trk and the master angle. Any suggestions?
Thanks
ak