Project Specs :

PC - MCA ver 3.5.4

24p PAL

Method2, 48kHz Wave

MXF 3:1, PCM

Video is digitized from Digital Betacam, 48k wave files are imported using method 2 and with BOX 1 and BOX 2 set to "OFF" in import options. Video and audio is sunk and spot check throughout the 1 hr long sequence shows they are in sync. Sequence is played with 24 fps play rate and crash recorded on DVD as well as Digital Betacam for sync checking. Usually in the second half of the sequence sync is lost between audio and video. Audio is advanced by 2-3 and in some instances 12 frames. The difference is not gradual but fixed throughout the scenes. Spot check shows they are actually in sync. To analyze this issue, I imported the same wave files in FCP and MC and got the following results.( In MC I used method 2 with both boxes set to "OFF" in import options) 

                # of Samples  Duration (FCP)  Duration(MC 24p)   Duration (MC 25p)    Duration (MC 1080/24p)

Wave001     1455199          30:08                 30:08                    30:08                       30:08

Wave002     7327679          2:32:16              2:32:17                 2:32:17                    2:32:16

Wave003     15229279        5:17:07              5:17:07                 5:17:07                     5:17:07

Wave004     7768319          2:41:20              2:41:22                 2:41:21                     2:41:21

Wave005    13284479         4:36:18               4:36:20                4:36:19                      4:36:19

Wave006     13719679         4:45:20              4:45:21                 4:45:21                     4:45:20

Why is there a discrepancy in duration between MC projects with different frame rates ? Even between 24p PAL and 1080/24p projects ? Could this be a bug and be the cause of the above problem ? Or could the problem be caused by audio not being locked to video during the shoot ? ( I don't have the information on how audio was recorded in the field). If audio wasn't locked to video during the shoot and sample rate for some scenes were slightly less than 48kHz although they were meant to be  48kHz and Avid stamped them as 48k then would audio run slightly faster in the timeline compared to video and cause the delay during those scenes ?Has anyone experienced similar problems ? I'll appreciate any feedback.

Regards,

Oner Olcerel

Keep in mind that a PAL 24p project is still a 25fps timecode since PAL is a 25 frame format. That would explain some of your discrepancies depending on what you were expecting project wise. Timecode from BWF files are calculated based on "samples from midnight." when imported you indicate frame rate used and then the START timecode is calculated. SD is 29.97 (DF or NDF) and PAL is 25. HD can have 24 and 23.976.

Michael

Thanks Michael. I still can't explain the inconsistency in the absolute duration of the clips between the projects. I would at least expect consistency between the same frame rate projects. I guess the question is how does MC deal with the residual samples of the wave file when it calculates the number of frames that would correspond to the wave file. Number of samples/48000 (for 48k files) will give us number of seconds plus a remainder. Dividing that remainder by 2000 (24fps) or 2002 (23.976fps) or 1920 (25fps) will give us number of frames plus a remainder. So how does MC deal with this last remainder ? FCP appears to be rounding down or up this remainder depending on whether it is less than or more than 1000 (for 24fps) thereby either truncates the file by removing the remainder samples or completes the frame by adding additional samples to the file. MC appears to be either adding additional samples or removing the remainder samples and in some cases adding so many samples that it amounts to more than 1 frame. As an example, in one case where the remainder was, say 800 samples, not only it added 800 samples to complete it to the whole frame but added an additional frame so the duration was 2 frames longer than the duration the FCP calculated.

Anyway, where all this originated from was the very major problem we have been experiencing lately. In the past we always had done the sync in TC. In this current project, we digitize video from Digital Betacam, mute, and import the 48k wave files for audio and sync them in MC. (Video is transferred from TC by 24 to 25 speed up). It is a 24p PAL project set up with "Film Rate". To make it simple, say they have 12 A/V clips of 5 min duration each after syncing the video with the imported wave files and create a 1 hr long sequence. The durations of the respective video and audio clips that make up the sequence are identical. Say there is a clapper at the start of each clip. When I do a spot check at each clapper along the timeline, the video and audio are in sych. At this stage, it is only a static check and not a dynamic one. The problem comes up when they do DVD dubs at the end of the day. The process is simple : Press play in the timeline and press record on the DVD recorder. So digital cut is not used. Play rate is set to 24 fps in the Film settings. I crash record on Digital Betacam at the same time to check the sync (I use embedded audio). Usually in the second part of the sequence, audio is advanced by 2 frames (In one case it was 12 frames). The delay is not gradual. When it was 12 frames out, say out of 12 scenes in the above example, the first few were in sync, next few were 2 frames out and the last one was 12 frames out. There doesn't seem to be a pattern in or logic behind this discrepancy. How would you expalin this discrepancy ? Could this possibly be explained by one of the following  scenarios ?                 

1. Video and audio are recorded unlocked to each other in the field (I don't have the details).

2. It is due to ver. 3.5.4.

This is the first time we're using 3.5.4. Should we try importing the wave files in MC 2.7.4 ?

I should point out that BOX 1 was checked and BOX 2 was unchecked in import options. As I understand, these options are not relevant in our case anyway.

Regards,

Oner Olcerel

Thanks for the detailed description! Can you do a quick test to see if you have the same playback sync issues when edit rate is set to 25 instead of 24?

Michael

I'll do that tomorrow Michael and let you know.

Oner

olcerel:

the residual samples of the wave file

I have not experienced the sync wandering off like you describe, not in 3.0, 3.1, 3.5.4, nor 4.0.2 and 4.0.3.

Because the wave files are not video-frame based, the total number of samples in the source wave files may not necessarily be and usually are not integral multiples of 1920 or 2000, leaving some "residual samples" when they are integrated into frame-based editing applications.

Ah, which recorder? I usually get stuff recorded on the Aaton Cantar, and that appears to complete to round seconds.

DEVA...

I've played out with 25fps edit rate (MC ver 3.5.4) and it was fine. However, I also downgraded 2 of the systems from PC ver 3.5.4 to PC ver 2.8.4 and it worked fine with the new audio imports as well. I also played out one of the previous sequences with sync issues by reimporting the wave files in ver. 2.8.4 and relinking them so as to eliminate possible wave import issues in ver 3.5.4 and it worked fine. I will do numerous tests using various combinations to obtain some consistent results. So far the delay issue seems to be related to a combination of ver. 3.5.4 and wave import in method2 and 24fps edit play rate however this is yet to be proved in order to arrive at a conclusion.

PS : The inconsistency in the durations of the imported wave files in 25p PAL, 24p PAL and 1080/24p projects is still haunting me. It may lead to different durations in 25 and 24 projects, but I would have thought the way MC deals with the "residual samples" would be consistent within the same project. Any interpretation on this ?

Oner Olcerel 

Michael, using Digital Cut to output the same sequence didn't exhibit the same issue. It appears to be only a timeline playback issue. It happened in a new project using the same media as well. I've tried it in 4 different systems using a combination of HP xw8000, HP xw8200, Adrenaline DNA, Mojo, v2.7.2 and v2.8.4. Where and whether sync jump occurs is not consistent. The above hardware, which meets Avids minimum requirements for the relevant versions should, I belive, support a long playback of a sequence @24fps in a 24p PAL project without any sync issues. Could the fact that video is captured and audio is imported from wave files have any relevance in Avid's handling of the video and audio when it is playing back the sequence in the timeline @24fps ? Although they are stamped to be sampled as 48k, could the real sampling rate of the wave files have any relevance in this issue ? Should the above hardware be able to cope with the the playback of such a sequence without video and audio losing sync ?

Oner

olcerel:

Could the fact that video is captured and audio is imported from wave files have any relevance in Avid's handling of the video and audio when it is playing back the sequence in the timeline @24fps

Well, don't want to say that it cannot have any influence whatsoever, but I (and at least five colleagues here in Amsterdam) have been doing this exact thing since at least 2002, importing sound from BWF's, syncing in Avid to captured Beta/DigiBeta/HDCAM footage, playing back timelines at 24.00fps and never had any issues. Not with Meridien, not with Mojo, not with Adrenaline, not with Symphony Nitris nor Nitris DX. This should work.

Thanks for that confirmation Job. It is very unlikely for all 4 different hardware not support it.

Oner

Job, we've tried the same project and media at another site and had the same issue. I've started to think whether it could be the codec that's causing this. It is 3:1 MXF and PCM(MXF) for video and audio respectively. I wonder whether MC is occasionally having problems when it is decompressing 3:1 MXF when playing back @24fps in a 24p PAL project. Have you ever used 3:1 in a 24p PAL Method 2 project and played the timeline @24fps edit rate for long enough without getting any sync issues ? What resolutions have you used ? Can you tell me how they create the BWF files ? Have you ever used DEVA ? Do the BWF files you get have the same TC as the video you capture and you auto sync or they don't have the same TC and you sync at the start ?

Oner Olcerel