Video cameras to timecode with Sound Devices MixPre II 10 for classical music videos

cupakm

New member
Hi there,

which quality video cams can you recommend for classical music videos? FullHD at 60 fps minimal, but 4K would be nicer. Need to keep them (at least 2 pcs) in sync with a Sound Devices MixPre 10 ii. Whole setup should be able to run on bats for 3 hours min. Small form factor would be a huge plus.

Best regards and many thanks in advance for your recommendations!
 
Is this for live, otherwise why do you need to keep them in sync, verses a way to get the sound in sync? You will most likely record the video with the camera with a scratch track for the sound, and capture the sound on a multi track interface to do the mix down before putting the two together.

I would speculate is you have someone narrating live, you would have the video recording in the background to pull the narration in and out, but the main mix would still come the multi interface for a better mix. Then the sync would then come from something like a clapper board so you can align the mixed sound to video sound. Hard snap, align video sound and mix sound, should be pretty close.

That is the way I would do it.
 
If they have a mic input you should be able to jam a time code into the audio track and have that used in post.

I like my older Panasonics but they don’t do 4K and not ready to spend the $ for all the storage and computer necessary to do multi cam editing of 4K clips so can’t recommend anything specific. The ones I have can run for a couple hours on the largest battery but they also work with external 5v packs (need to get ones with sufficient amp capacity) and could go forever. 4K will gobble SD space at a good clip and I expect battery life too.

Video is an expensive hobby to add to audio. I found it not so satisfying because you really need eyes on the cameras and that’s hard to do unless there’s good phone apps for your choice of video recorder.
 
Then the sync would then come from something like a clapper board so you can align the mixed sound to video sound. Hard snap, align video sound and mix sound, should be pretty close.

It's close, but I find it's not close enough. In my experience it's necessary to fine tune the sync by hand to get it really looking right. On my cameras (four consumer grade Sony digital camcorders and three older GoPros) the audio alignment to the video track varies by two or three frames, which is more than enough to make the sound seem just a little off for music. For these same reasons, software that syncs video by the audio tracks just gets it close and it needs manual fine adjustment. Time code on the audio track would have the same problem.

In the screenshot below you can see three audio tracks. The light colored one is the final audio carefully synced to the video. The shaded tracks are unused camera audio from video that has been perfectly matched by eye using cymbal hits etc. The selection on the timeline is two frames. The difference in the waveform is from the cameras picking up the guitar amp from different places in the room.

audio vs. video sync.png
 
It's close, but I find it's not close enough. In my experience it's necessary to fine tune the sync by hand to get it really looking right. On my cameras (four consumer grade Sony digital camcorders and three older GoPros) the audio alignment to the video track varies by two or three frames, which is more than enough to make the sound seem just a little off for music. For these same reasons, software that syncs video by the audio tracks just gets it close and it needs manual fine adjustment. Time code on the audio track would have the same problem.

In the screenshot below you can see three audio tracks. The light colored one is the final audio carefully synced to the video. The shaded tracks are unused camera audio from video that has been perfectly matched by eye using cymbal hits etc. The selection on the timeline is two frames. The difference in the waveform is from the cameras picking up the guitar amp from different places in the room.

View attachment 106086

Right, wouldn't a snap with a clap board or some sharp noise get all of the audio sources so they could be aligned? That was what I was thinking, have the clap as a reference point to align. I wasn't even think automatic. I guess I stt do things the hard way.
 
Right, wouldn't a snap with a clap board or some sharp noise get all of the audio sources so they could be aligned? That was what I was thinking, have the clap as a reference point to align. I wasn't even think automatic. I guess I stt do things the hard way.

Absolutely, but the alignment of the audio tracks to their own video tracks varies by two or three frames between cameras so you still have to fine tune the timing of the video tracks.

If you're planning on using the audio from multiple cameras and there's enough in common to use a clapper board then there's enough in common that phase mismatch might be an issue.

I use a clapper board, but I generally rough in the video sync manually using the audio tracks then perfect it by comparing drum hits or guitarist strumming motion. Similar with the final audio, I'll drop it in and line it up to one of the camera audio tracks then nudge it around until it looks right.

Another nice way to line up the video is with a camera flash. I've done that but even then it can be off by one frame.
 
That is the way I would do it.
Hi DM60, thanks for your answer. Yes that is the way I was and still am doing it.
They are classical music videos without any narration. I just wanted to speed up the post, because it consumes time to adjust each take from each cam with mastered audio. So I thought I could skip all those steps and have just audio switched for the master and that would be it.
 
Using a clapper might work but for a 3 hour shoot, you could end up with some drift if all the clocks are accurate. In addition, you have to keep all cameras running the whole time. You can't stop and start unless you do a snap each time.

If you're going to do it for real you can get a system like Tentacle Time Code sync. You buy units, sync up the timecode clocks and insert that into either an audio track on the camera, or sync up to the audio recorder. They have software that will move everything into since based on the the timecodes.

Its not a cheap system but you can use DSLR cameras instead of shelling out for professional grade cameras. Its expandable, so you can sync up as many devices as you need.

YouTube
 
Using a clapper might work but for a 3 hour shoot, you could end up with some drift if all the clocks are accurate. In addition, you have to keep all cameras running the whole time. You can't stop and start unless you do a snap each time.

I assume more recent DSLR and mirrorless cameras have overcome the limited record time of the earlier DSLRs I've used. I think it was about eight minutes, perhaps a 4GB file size, then you had to restart the camera.

If you're going to do it for real you can get a system like Tentacle Time Code sync. You buy units, sync up the timecode clocks and insert that into either an audio track on the camera, or sync up to the audio recorder. They have software that will move everything into since based on the the timecodes.

Its not a cheap system but you can use DSLR cameras instead of shelling out for professional grade cameras. Its expandable, so you can sync up as many devices as you need.

I would expect that system to have the same 2-3 frame (4-6 @ 60fps) variation between cameras since it relies on the same audio track. But at least it gets everything close enough that it's just a nudge to get things perfect.

Also, how does it work out if the clocks are different enough that different cameras drift by a couple of frames over a three hour shoot? That said, even my consumer grade gear seems to be good for at least an hour. That includes Sony camcorders, GoPros, Zoom H5, Alesis HD24.
 
I suspect that the clocks are going to be extremely close, being all the same design, vs 3 or 4 different designs (or clock chips). I used to check stopwatches by starting them vs the Atomic Clock, and then letting them run over the weekend (for ISO calibration purposes). They were routinely within 1/10 second over a 3 day period. We had one brand that was off by a second or more over that period. I could buy NIST calibrated stopwatches for 4 times the price, but it was easier to buy 10, check them out and toss the 1 or 2 that didn't match up.

The other nice feature is the software that will move all the video fragments to equal time code spots. If you have 20 or 30 fragments each from three different cameras, that's a lot of manual nudging with little to use to manual sync. I haven't used the Tentacle system myself, but I can see where it would beat my normal method of trying to use a cymbal strike to align an audio and video. I've done that quite a few times, and sometime I just miss it by a hair. Bugs the crap out of me, but since I'm just doing it for fun, its fine.

If I was charging money for this, I would want it to be spot on.
 
Even with the Tentacle system I would want to manually fine tune the sync. I have two Sony camcorders of the same model and their audio tracks can be off by a couple of frames. I suspect it's just a result of when the start button is pushed and the video sampling begins. I have a third Sony that's essentially identical but one level up in features, and the fourth that's a different style but still falls within the same range. Even the three GoPros vary despite two being identical. I think it's just quantization of the frame capture vs. the much higher sample rate and lower bitrate of audio.
 
Why wouldn't the audio be more that sufficient to sync video which would be only 1/30 or 1/60 of a second per frame. That's low bandwidth compared to audio. Since the audio track in synced to the video track in that camera and is generated by the Tentacle, it should maintain alignment. The only thing is that the Tentacle has to remain plugged in, and uses one of the audio tracks. 3 cameras need 3 Tentacles. According to their info, the devices are good to <1 frame/day drift.

You end up with 3 audio tracks that are synced to 1/100th of a second. I can't sync that well. I was looking at my Powerdirector software and apparently it can do audio sync for multicamera work. I've never tried to use that.... maybe this summer if we ever get to leave the house. I might already have a couple of video files that I can test. I need to look through my files. I would be a nice thing to test.

If it wasn't about $250 per device, I might try them out. Its a bit pricey for a pure hobbyist.

As for the OP, that system would open up the choices on possible cameras to use. It should work with any camcorder or DSLR that accepts an external audio signal.
 
Why wouldn't the audio be more that sufficient to sync video which would be only 1/30 or 1/60 of a second per frame. That's low bandwidth compared to audio. Since the audio track in synced to the video track in that camera and is generated by the Tentacle, it should maintain alignment.

Because the audio tracks will be offset from their video by as much as 2-3 frames (@ 30FPS), and that offset will not be consistent between cameras, even identical cameras. Yes, the offset will remain stable on one camera for one recording but it will be offset. It's not drift, it's a consistent offset.

Video may be fewer frames per second than than samples per second of audio, but it's a lot more data per second. I suspect there's a buffer that stores several frames of data before it gets written. I suspect the variation is a matter of where in the frame capture/write cycle the camera is in when you hit the record button. Audio starts recording almost instantly while the video waits to capture the next full frame to the buffer before writing.
 
I'm amazed guys!

Sync is a NON-ISSUE30 years ago, it was a constant pain and a battle - SMPTE lock ups to loads of drifty unreliable analogue gear. Now, if you have a mix of modern card recording domestic, prosumer and pro gear, then as long as they are all locked to the same frame rate you can sync your cameras at the start of a one hour recording and forget it. You can develop bad habits and they cause ZERO sync issues. Drift and sync loss is VERY VERY rare.

It doesn't matter if your camera produces a single long file, or multiple short ones. Butt them together in the editor and sync is solid.

Depending on the music, you could use clever software to do the sync, but I haven't needed that for years - plural eyes worked well, but just not needed.

The key factor is the audio waveform in your editor. It's a roadmap to getting the cameras close, and it's very obvious. with the entire recording visible, slip and slide roughly because the start time of each camera is different. Then find an event. The big tutti section, or if you remembered, you standing in view of every camera clapping, or that obvious cymbal crash. If it's prominent, then every camera's mic will have captured it. Zoom in on that in the waveform, and slip, frame by frame till they match. Remember the distant camera will be audio behind, and at the frame limit, you might be unable to get the peaks to align. I just allow the further cameras to be audio behind rather than in front. In Premiere I then trim the beginning of every camera to align. That's sync done.

Problems? I have one camera, about ten years old now that has a real fault. Randomly, maybe once every five-ten minutes, it drops a frame. I know this, so before I start editing, I run through the timeline looking for mismatched audio peaks from this camera. I cut the track and drag the timeline a frame, and it's back in sync. I usually just extend the left side to close the gap, effectively duplicating the odd frame. If one of these happens to be visible in the edit that follows, I just fix it with a one frame shift then.

Camera audio allows this to be totally foolproof. The only time I got caught out was in a large church where I managed to get one camera right at the very back, and the audio took 200ms or so to get there, and I forgot to compensate as the result was a delay on the video. The conductor spotted the batton wasn't quite right. I hadn't realised they could spot their hand actions being out of sync by such a tiny amount!

Churches, choirs and large events are not at all problems for sync. With gopros, I can have 8 streams to sync on stage, and it's never an issue.

I just finished recording a studio recording where sync was trickier. People were comping vocal and instrument tracks in two bar phrases. The video files would have between two and six attempts at each phrase. They then comped these takes into the finished product. I was a little worried about the sync, because they did the usual audition each one and move it to a new comp track, so I didn't have any information on which of the takes on the camera was the selected one. However, once I got the completed comped track, visually there were differences between each take so I'd look at the take in Premiere, and see maybe 4 waveforms, I'd have underneath the comped track, and then with a little slip and slide, the right one was fairly easy to find, and sync - and even when two looked exactly the same on the screen, even the wrong one synched. I think I can spot a couple as I'd heard the damn things over and over again, but nobody I tested them on could! For interest, this does fall down on BVs - where they use time stretch to align different takes. Not sure if you've come across this, but imagine maybe five people singing their lines, but with the way they say the wordds and the timing slightly different. Cubase, for example, lets you place them abive each other, designate one as master, and then the others get stretched and squashed to fit. Sounds great, but messes the video up no end. Luckily in the video, BVs where there are multiple voices don't seem to look wrong. You'd spot it in the lead vocal, but it's OK in the backup vocals.

Just don't get spooked by video sync - In Cubase we fiddle with teeny timing issues. In video we're locked to video frames. For me, it is a very rare thing to have to take video into Cubase to shift audio by a quarter or half a frame. I can do it, but it really doesn't matter with live recordings because audio in an orchestra/choir/on stage always has very large delays as part of the experience - 300 Metres = 1 second is quite a long time when we so often moan about 10mS latency being intollerable!!
 
I'm amazed guys!

Sync is a NON-ISSUE...

Drift and sync loss is VERY VERY rare.

Drift isn't the problem. Using audio leads to offsets of a frame or three, which I can see. And the audio tends to be early by that much, which is much more unnatural than late audio.

It doesn't matter if your camera produces a single long file, or multiple short ones. Butt them together in the editor and sync is solid.

Well, no. The AVCHD files produced by my cheap Sony cameras have a 2-frame gap between them. Fortunately those cameras will simultaneously record MP4 files that cover longer times and I can steal two frames to fill that gap. My two old GoPros make files that have an overlap of approximately 1.5 frames. To get those to sync up on longer takes means I have to alternate between overlapping them one frame then two frames and so on.

I can do it, but it really doesn't matter with live recordings because audio in an orchestra/choir/on stage always has very large delays as part of the experience - 300 Metres = 1 second is quite a long time when we so often moan about 10mS latency being intollerable!!

I shoot rock and country bands. Their timing is inherently tighter because they don't have the physical separation you get from an orchestra. In audio a timing difference of 25ms is about where things transition from sounding like phase interaction to sounding like a doubling. It's an audible interval, and it's less than a single frame at 30fps. It's also a visible difference. So the audio is sufficient to rough in the alignment of the video tracks, but to make it really look right I have to nudge them around up to three or four frames.

Since the camera audio tends to be on the early side I also need to place the finished audio later on the timeline. In the video I just finished the mixed audio is 67ms later than the earliest camera audio track.
 
I don't get any of those issues. My Panasonic AVCHD camera is totally fine with the files joined in the editor. With the bands, a frame at the Uk 25fps slid forward or backward when the real sync point isn't critical for me at all. I've got some of the cheap Chinese Gopros for bands, and they're 30fps, not 25 so it means a frame rate conversion and the audio lines up fine. We normally take the audio as multi-channel from our M32 and X32s - the mix gets dumped as stereo into the editor and a drum downbeat works great to align them. The only repeated sync during the gig comes from the batteries dying, and needing to be sorted during the show - so we might have 5 minutes of nothing, which then needs resyncing.

Why do you say a frame or three for the audio - I just get them accurate on a single frame. If the camera is on something like the drums this I agree might mean slightly early is better than slightly late - but frame accuracy is OK for me with our JVC cameras that record in mp4 or .mov format normally, and the smaller AVCHD ones. I've still got my old Betacams in the store and I could set the timecode on the JVCs, but I never found a need for it.

Nudging 3 or 4 frame from the audio waveform on the screen would worry me. I just make sure the audio waveforms align and that's never an issue. The only exception is with a distant camera and long lens, when the audio is late , but once I sort that one, I'm done.
 
Talking about 10 or 20ms delay, that would just about cover things if your camera was 10 or 20 ft from the band. If you were taking the audio directly from mics on stage and/or the PA mix, then having a camera 20 ft away should generate a delay of 20ms based on the speed of sound. That's where an external timecode would work better than trying to sync audio from the camera.

Most of my video work has been fairly close, 10 ft or less. Of course there's no phasing or doubling issue, since the audio track from the camera is tossed anyway. Only the audio from the Zoom makes it onto the video.
 
Nudging 3 or 4 frame from the audio waveform on the screen would worry me. I just make sure the audio waveforms align and that's never an issue.

Why would that worry you? Neither of us are using camera audio for anything but sync so who cares if the camera audio is off? I do my final sync on the video itself anyway. If I aligned by audio only different angles would be visibly off from each other, which would be an issue.

Here's what the audio files in my most recent project looked like after I synced the video. The top audio track is the audio mix actually used and the rest are camera audio tracks not used. Since I had "Quantize to frames" on, the video tracks could only move by a full frame. If I had placed them where the audio matched best the video tracks could have been as much as two frames off, which is visible to me. If I had placed the audio mix using the camera audio as a reference it would have been noticeably off.

Maybe I'm just more sensitive. I can hear the delay added by my Bluetooth speaker compared to the onboard headphone output or my USB interface.

audio-video sync.png
 
Last edited:
Ironically one of my old clients sent me a pile of clips to edit today and instead of complete song length tracks, I've got a couple of bars of a solo, then a verse guitar bit, then some twiddlybits from a guitar less than a bar long - plus the end to end audio track is lacking the vocals which would help no end. This is one example where time code would have made the thing so easy. Words, and eat, come to mind. Trying to find the right place in the song where the visuals come from is agony!
 
Back
Top