PANASONIC AG-DVX: Panasonic AG-DVX Forum Panasonic HVX - HPX Panasonic Varicam

TC problems with FCP 3 and DVX-100

COW Forums : Panasonic AG-DVX100

Return to posts index   •   Read entire thread

Patrick O'Connor
TC problems with FCP 3 and DVX-100
on Oct 7, 2003 at 12:04:28 am

This is really a post-mortem discussion of a project that's already done.

First, disclaimers:

1. I shot edited and post-supervisede a film for a friend who was commissioned by the IFP (along with several filmmakers) to shoot a short (5mins) piece to be shown on Windows Media 9 Cinema Projection, using the DVX-100. We shot in standard 24p mode (not advanced) on the DVX100, used the camera for logging/digitizing via Firewire into FCP 3. The final edits were to be conformed to 23.976 fps using a new Quantel edit system and Panasonic DVCPro deck.

2. We were not giving the right information up front (not the IFP's fault). Basically, all of us digitized from the DVX100 and edited in FCP 3, instead of 4, which could have created some problems. As far as I understand, Panasonic and/or Apple decided not to bother with support for 3, focusing instead on support for 4. This fact alone could have accounted for many of the problems we encountered.

The most basic problems I enountered were that during capturing on the DVX100 via FCP 3, I would get repeated messages about dropped frames and timecode breaks. When I contacted our tech person on this, he said not to worry about it - as long as the footage was captured in FCP and there were no visible dropped frames in the timeine, we could basically export a targe sequence of the entire project, which we'd import into Quantel and then pull out the interlaced frames . So, ahead we went. Everything was fine after initial capture problems - i.e. the footage in Final Cut appeared okay -no synch problems, no dropped fields, etc (even though some had been reported).

When it came time to conform on the Quantel, we had huge issues. The post flow worked like this: from our FCP timelines we created EDLs. These EDLs were convereted to Advanced Authoring Format (.AAF) in the Quantel system - this utility converted our 29.97 EDLs to 23.976 EDLs. These AAF files were imported to Quantel, and we recaptured. [We couldn't export directly to AAF format because FCP 3 doesn't support it at all, and FCP 4 has not yet added true support for AAF (if I understand this correctly, AAF support will be part of the XML implementation, which hadn't occurred when we did this (two weeks ago)]. I'm not sure if it's the Quantel system or the Panasonic deck used for the capture, but the interlaced frames are automatically dropped at this point - they come into the system as ture 24p, rather than as 29.97 with 3:2 pulldown added. I think the camera labels the interlaced field during shooting, and the deck, when set to 23.976 mode, ignores the labelled fields on playback.

Tape 1 was fine - no issues. Tape 2 was a disaster - timecode offsets all over the place. Non consistent, either - if shot 1 was off by 1 min and 12 frames, shot 24 could be off by 15 seconds and 4 frames - so we couldn't just recalibrate TC for that tape and move on. The only differences using tape 1 and tape 2 is that I did NOT use the TC Recheck button on the camera for tape 1, but I did use it on tape 2. It's interesting that the very function intended to prevent timecode breaks on the camera seemed to completely screw up the TC - this is a theory, not a statement of fact, but it's the only difference between the two tapes that I could think of - all other settings were the same.

So we tried to use the Targa sequence method - exporting all frames of the timeline as a sequence of Targa files. This did not work however, as there seemed to be an issue with the field order (upper or lower). Quantel wanted upper field order, Final Cut defaults to lower field. I was not able to figure this out, even after exporting two Targa sequences - one from a timeline's field order was set to lower (this shouldn't have worked), one from a timeline's field order that was set to upper (this should have worked but didn't). I suspect that at some point during the original capturing phase, we were stuck with lower-field first footage, and that switching to a timeline with a different field order was not going to work.

So we had to go back to the timecode problem. We were able to identify where this problem was occurring. For some reason, the TC coming from Firewire was different than the TC coming over Serial device control. The filmmakers all used Firewire for logging/digitizing, the Quantel uses serial device control. The solution (in my case), was to import the EDL into Media 100 hooked up via serial device control to a Panasonic DVCpro deck, relog and digitize tape 2, and then manually correct the shots from tape 2. I exported an EDL and this seemed to convert correctly. This took about 2 days. This happened to some degree, as far as I know, to all of the filmmakers. I'm not an expert, but apparently the serial control sends LTC, and Firewire control sends VITC, and these were out of synch. To test the theory, I put the same tape, stopped on the same frame, in two different decks: Panasonic DVCPro (an ancestor of the current AJ-D455) and a Sony DSR20. The timecode displays for the same frame on the picture were different on each deck. Possibility: The DSR20 was reading VITC and the Panasonic was reading LTC. NOTE: I am not a professional with VTRs, so I could be totally wrong here - but suffice it to say that one deck gave me timecode x and the other gave me timecode y.

We also had significant problems using FCP 3 EDLs with Quantel AAF conversion tool. Basically, any audio events in the EDL wreaked havoc, especially because we had all used a lot of audio from different takes to fill in the soundtrack (i.e., room tone, ambience, etc.) Having "solo" unsynched audio items on the timeline seemed to completely confuse EDLs. The solution was to export only the picture edits, and then export the audio portion of the timeline as a WAV file, and import that into Quantel. This is not surprising to me, however, as FCP 3 EDLs can be notoriously problematic for conforming for negative cutters or Avid editors. NOTE: The original FCP 3 EDL worked fine imported into Media 100 7.5.3, and both EDLs, for all intents and purposes, matched.

We also had issues getting a title sequence in from After Effects, but this somehow worked (probably by accident), with one of the targa sequences we tried (if I remember correctly, we used Boris Titling plug-in in AE, and exported the sequence at 720p resolution at 24 fps with motion blur enabled). It seemed to working Quantel quite well.

What are the lessons?

If you're conforming to 23.976 or 24 at a later date...
The first is that you should be sure to use FCP 4 for capturing and digitizing - and make sure you've done your research (this forum is a great place to find info). The second safety procedure is that you ought to try to keep the device control path consistent if you are going to reconform to 23.976 on a system like Quantel. I.e., if the Quantel is reading serial device control, log and capture with serial device control. Unfortunately, that's not possible with the DVX-100 - you'd have to find a deck that is using serial device control, even if you are still capturing picture an audio over Firewire.

If I had known this in the beginning, it would have saved me days of wasted work.

I also welcome comments, suggestions, workarounds, from anyone on this particular test case. While I have a lot of experience with film matchback, I am brand new to 24p (either SD or HD) and would really benefit from the opininions of those who could take my theories and validate or disprove them.

Also, it is worth noting that I don't believe that any of the systems listed above are faulty by design. The systems by themselves are great - the Quantel is amazing, the DVX-100 is hard to put into words (hands-down the best image I've seen in any <$3k prosumer camcorder), and Final Cut is of course my system of choice. It's just that putting these together using this post workflow is fraught with gotchas.

I am still confused, however, by the disparate timecodes on the same tape - I assume this is a DVX100 problem, but I'm open to other options.

One last thing to note - some of the filmmakers had footage that came in as 29.97 NDF (all colons - my project was one of these) - others had standard DV timecode - 29.97 DF. My project did not mix and match - it always came in as 29.97 NDF, Firewire or Serial, on the DVX 100 or on any deck.

Thanks Patrick

Posts IndexRead Thread

Current Message Thread:

© 2018 All Rights Reserved