Audio and Video Question

[saralikestea]

Hello,
I was wondering if professional filmmakers record audio and video at separate times. If so, how do they make the sound and picture line up?
Sorry if this is a stupid question, I'm very new to this!
Thanks so much!

[Cornucopia]

Sometimes yes, sometimes no.

Even when they record it at the same time, they rarely if ever will record both picture & sound to the same device ("single system"). Usually, they will record high quality picture & med. quality guide/reference sound to the camera's storage (whether celluloid film, video tape, or digital drive) and then they will also record high quality sound to a dedicated sound recorder (aka "double system").
When setups are fully high-end professional, both cam & sound recorder are sent generated sync clock & timecode (from same master source), aka genlocked. Often this is coming from the slate/clapper.
Even if they are more "wild", they still will start (and sometimes end) EACH of their takes with the slate/clapper so there will be an easy to find, distinct mark on the recording, both visually and audibly.

Once in the edit studio with the clips loaded in to the NLE (one that allows full/fine audio editing as well as video), they will sync via that same clapper mark. Plus if you sync only via sound vs. the ref guide sound, you should easily see the waveforms (nearly) matching their onset/rise, especially at the clapper marks.
Assuming there is no clock drift, they should line up for the duration of the take(s).

If you are recording before the fact (as in music videos), they just have a genlocked location playback source & match ("jam sync") the timecodes.

If you are recording after the fact (as in overdubbing, ADR - automatic dialogue replacement), they have a genlocked studio playback monitor and use a pro DAW setup like ProTools where you can be given a countdown beep tones/light flashes to be ready to go into record. Then they practice, practice, practice and loop, loop, loop record so they end up with multiple takes to where they pick the closest/best take. There are also special plugin tools that, using the timecode, the clapper marks, and the sound waves, will attempt to automatically simplify the syncing to where it's close enough to just tweak/finesse for the final version. This includes both timeshifting, time compression/expansion, micro-edits, and waveform morphing/wapring.

It's just smart practice to use Timecode and Clapper/Slates if you ever expect to do this professionally, especially when those can be utilized cheaply on a smartphone/tablet now.

Scott

[aedipuss]

lol yes, it's all part of learning how to edit video. audio is a large part of making videos nice....

[JVRaines]

I would like to commend Scott for such a correct, patient, and thorough answer. This is VideoHelp at its best.

[johnmeyer]

+1 to Scott's amazing answer.

[lingyi]

Once in a great while, especially if someone uses their hands for the sync instead of a clapper, I've heard/seen a slight audio/video mis-sync that isn't noticeable in normal speech/action on the show.

I don't know how common this in other countries, but I've mostly seen it (the clapping of hands, not necessarily the mis-sync) on Korean variety and reality shows as they sometimes show what happens during a tape change or return from or entry into a break. This is usually done as an extra bit from the guest(s).

[Sharc]

A certain delay of audio vs video is just natural due to propagation delay of audio which is for example 20ms for a distance of about 20 feet.

[Cornucopia]

While you are right in the calculation about the amount of delay at that distance, I have 2 points of clarification (rebuttal?) on that topic:

1. Running at 60pFPS, this delay of 20msec amounts to 1 video frame, which has been found to not be noticeable by a vast majority of people. The minimum threshold for most is ~40-45msec. To be obvious, one has to get beyond 100msec and up to ~240.

2. 20feet!? If they are shooting and holding the microphone THAT far away, they are terribly unprofessional and doing a major disservice to the quality of audio, regardless. Particularly, when it comes to sound isolation from ambience - I would expect a wireless lav to be IN CONTACT with the talent, and a boomed shotgun should be somewhere between 0.5-3.0 feet from talent, similarly with other forms of hidden mikes.
So the delay you SHOULD be experiencing should be max 20/3 or 6.6msec. Very minor indeed.

My conclusion must be that for those clips you refer to, for there to be noticeable mis-sync, somebody doesn't know how to shoot, edit, and/or encode/distribute properly.

Scott

[saralikestea]

Thank you so much for the explanation!

[Sharc]

While you are right in the calculation about the amount of delay at that distance, I have 2 points of clarification (rebuttal?) on that topic:

1. Running at 60pFPS, this delay of 20msec amounts to 1 video frame, which has been found to not be noticeable by a vast majority of people. The minimum threshold for most is ~40-45msec. To be obvious, one has to get beyond 100msec and up to ~240.

2. 20feet!? If they are shooting and holding the microphone THAT far away, they are terribly unprofessional and doing a major disservice to the quality of audio, regardless. Particularly, when it comes to sound isolation from ambience - I would expect a wireless lav to be IN CONTACT with the talent, and a boomed shotgun should be somewhere between 0.5-3.0 feet from talent, similarly with other forms of hidden mikes.
So the delay you SHOULD be experiencing should be max 20/3 or 6.6msec. Very minor indeed.

My conclusion must be that for those clips you refer to, for there to be noticeable mis-sync, somebody doesn't know how to shoot, edit, and/or encode/distribute properly.

Scott

I agree with your comments. I mentioned the delay just because it is wrong to conclude that video/audio are out of sync when one sees in an NLE (for example) the audio waveform lagging an event happening in some (far) distance, and trying to correct it.

[Cornucopia]

I agree. However, if there is a static relationship between the capture distances & thus also the media event times, I WOULD correct for that as long as that didn't affect some other physical-based-assumed context. Minds usually do prefer things to be in sync.

Scott

[johnmeyer]

I do these audio delay distance calculations all the time for filming stage productions. One that I did for several years at a high school gym, was on a stage at the other end of a basketball court. The speakers were set up on the stage at the end of the court, and I was on the other end. People on stage were mic'd. The delay over the distance of the length of a basketball court was three NTSC frames of video, enough to be slightly noticeable. For larger auditoriums and outdoor venues, I've sometimes had to make larger corrections.

I always get a direct feed from the soundboard, and getting this sync done becomes even more important because I fade between the soundboard (which sounds crystal clear, but devoid of any ambience) and the ambient sound in order to get crowd reactions and make the audio sound real.

[Cornucopia]

Yeah, Been there. That distance, I would expect issues.
And dealing with combinations of close + distant sound can screw with phasing/eq too.

Scott

[johnmeyer]

Yeah, Been there. That distance, I would expect issues.
And dealing with combinations of close + distant sound can screw with phasing/eq too.

Scott

I can't hear or see writing about phasing without thinking about the first time I heard it, during the drum solo on the 1968 Iron Butterfly classic "In-A-Gadda-Da-Vida." I had no idea what it was, but it sounded great. However, when you don't want it, phasing and its cousin, flanging, are pretty awful to listen to.