Half a frame audio drift from 3 cam interview shoot - any ideas why?

Hi guys!

I'm new here, so thanks for this.

Over the past few months, I've shot 6 multicam interviews.

They were shot with 3 x GH4's, with audio being fed into the 2 cams taking the single shots of the interviewer and guest, via Rode Link wireless kits.

The wide shot just uses on board audio as a scratch.

The first 5 interviews synced up via FCX's audio sync when creating the multicam clip - no problem - in FCX 10.2.x

The 6th interview does sync up, but with both the interviewer's and the guest's audio turned on, I could instantly tell there was a slight sync issue. Now I'm using FCX 10.3.x

I could hear the interviewer in the guest's mic a tiny fraction after the interviewer spoke, so it gave it this weird stereo effect.

Upon inspection the audio from the 2 rode links are about half a frame out of sync.

Now, I couldn't move the footage in the multicam timeline, as the issue is less than a frame, and I can only move the video a frame at a time.

Instead, I nudged the audio up half a frame on one of the tracks BEFORE creating the multicam clip.

Now this works perfectly, but after 50 minutes or so, I start to see a fraction of a frame of drift again.

It's not horrendous, but hasn't happened on those first 5 shoots, so I'm trying to figure what could have caused this?

All settings on the cameras are identical on ALL 6 shoots.

The project settings in FCX, fps settings and 48khz audio etc, are all identical too.

Things I could think of:

- Lav mic placement of guest did look a little less directional to the guest than on other shoots. Dunno if this would pick up the interviewer more at times?

Would the time difference of sound entering the mic cause this kind of thing?

- Final Cut 10.3.x processes the audio differently to 10.2.x and this has caused a problem with the sync?

- One camera dropped half a frame randomly? Dunno if that ever happens, or is possible?

- I'm being unrealistic that the sync should be spot on every time?

As I said, when I looked at the sync in the multicam timeline it appeared that FCX couldn't possibly sync the 2 video clips perfectly, as to do so, would have required moving one of the videos less than a frame.

Video can only be moved a frame at a time, so is this process always going to be flawed?

How can you ensure that the audio sync from 2 cameras doesn't fall between frames?

I told you I was new to this, I'd love some education on how all of this works, and whether perfect sync is ever actually possible?

Thanks!

ross.grant wrote: ...multicam interviews...I nudged the audio up half a frame on one of the tracks BEFORE creating the multicam clip...this works perfectly, but after 50 minutes or so, I start to see a fraction of a frame of drift again....hasn't happened on those first 5 shoots, so I'm trying to figure what could have caused this?...All settings on the cameras are identical on ALL 6 shoots...

Using normal consumer equipment you can't rely on long term audio sync. Some of the reasons were discussed in this thread: www.fcp.co/forum/4-final-cut-pro-x-fcpx/...ery-long-shots#85550

The question is not why did the sync drift, but how could it have possibly stayed in sync before? It could have been almost anything: varying temperature, etc. The crystal oscillators in each camera that ultimately control recording speed are not temperature controlled, hence can drift based on ambient conditions or even aging. I doubt they are individually calibrated.

This is normally not an issue with scripted narratives, documentary, commercials, weddings, tutorials or most interviews. The average take length is quite small. However with long interviews or classroom-style instructional content, this can definitely happen. You can't predict how the cameras or audio recorders will behave.

One option would be to sync with PluralEyes which corrects for sync drift: www.redgiant.com/products/pluraleyes/

Another option is to use external hardware timecode on each camera. This also allows easy sync if cameras are stop/started many times. The least expensive option is Tentacle: www.tentaclesync.com/

… and another option is to not mix two mics.

half a frame isn't noticeable as a lag btw. audio and video; but very noticeable as some kind of reverb/flanging - IF both mics are on.

So, when the interviewer is talking, mute the other mic and vice versa.
Hopefully, it isn't a lively debate where all chat simultaneously :woohoo:

Thanks so much for this info.

I have Pluralwyes, so will see how that fairs.

Also, I notice that the GH4 can have it's time code synced via the mobile app.

I have used this to sync all 3 cameras to the current time as timecode.

Has anyone tried this on a shoot, and if so, how accurate is it?

Would tentacle ensure perfect sync every time?

I'm happy to spend the money if so. Thanks!

This won't fix things now, but for future reference, since you only have two mins, put both mics on one camera, avoid the headaches. We have a firm rule that we record to camera, and to as few cameras as physically possible. This has knocked out a lot of problems we used to have.

Unfortunately I'm using Rode Wireless Link kits, so you can only run one receiver to one transmitter - hence I record the audio to each camera recording the singles.

I just realised too - if those wireless kits are plugging in via the 3.5mm jack - then that means the Tentacle units won't have any where to plug in?

ross.grant wrote: ...Also, I notice that the GH4 can have it's time code synced via the mobile app...I have used this to sync all 3 cameras to the current time as timecode...Has anyone tried this on a shoot, and if so, how accurate is it?...Would tentacle ensure perfect sync every time?...

I think that just "jam syncs" the three GH4s one time, assuming the cameras are all set to free run timecode. That is a nice method, and allows multiple cameras to start/stop yet maintain sync. Dave Dugdale used a similar technique in this video:

My crew doesn't use that because our diverse cameras don't all support jam sync via IR or remote command. If they did I would definitely use it. It also doesn't work for audio recorders.

However it does not ensure long term freedom from sync drift. Your best option is probably try PluralEyes, which is also much more capable at audio sync of difficult cases than FCPX. One time we shot an entire multi-day project without having the audio recorders set to the correct time of day, which would have made manual sync impossible -- it was thousands of video clips, yet PluralEyes matched them up by waveform.

I haven't used Tentacle but it is the least expensive way using external timecode. It apparently uses a high-precision oscillator and won't drift more than one frame per 24 hr. It can also handle audio recorders which don't support SMPTE timecode, such as the Zoom H4 and many Tascam recorders. Those display a pseudo timecode but it's not SMPTE and there is no timecode metadata recorded. Tentacle can record an audio timecode on one stereo track and their post software can convert that to a format FCPX can load (I think).

The downside to external timecode (besides the expense) is it's an extra item per camera so it increases complexity. I personally would rather take my chances with PluralEyes. However there are cases where Tentacle would really help -- say you have several roving shooters at an event, they aren't in range of common audio, they are all doing stop/start yet you want to sync all those cameras in post.

In your case you're shooting interviews so you have good common audio -- it's not like you have multiple cameras on either side of an auditorium. I would tend to try PluralEyes first not Tentacle.

You can get that mini jack split into left & right channels.

Have you actually verified sample rates being recorded?