[time-nuts] Re: Where do people get the time?

[John Hawkinson]

Hi. (former) Movie projectionist here. Some comments on multiple messages inline below:

Hal Murray <halmurray at sonic.net> wrote on Sun,  2 Jan 2022
at 12:57:24 EST in <20220102175724.6D8AB28C19B at 107-137-68-211.lightspeed.sntcca.sbcglobal.net>:

> How far off does the audio have to be before it doesn't look/sound
> right?

Although this might be an interesting question, it hardly matters in practice. Picture and sound were always recorded seperately in professional movie contexts (indeed, the photochemical process and magnetic tape processes didn't lend themselves to sharing a single recording medium). Although nowadays cameras can and do record audio, with multiple microphones audio is (usually) separately recorded. So it's always been a matter of syncing them up.

> How accurately does the typical movie process get things
> aligned?

Traditionally film was 24 fps so getting sound to match up with a single frame was the best you could so, and wasn't hard to do. So that's 41.6 milliseconds. Indeed, a bigger problem is the effort that was put in to make sure that camera and sound recording equipment were frequency locked to each other so there was no *drift* over time.


Lux, Jim <jim at luxfamily.com> wrote on Sun,  2 Jan 2022
at 15:17:30 EST in <ab4f9a1a-4313-b0d7-4ed5-00fc0ea00785 at luxfamily.com>:

> I don't know today, but back in the day, probably 1/24th of a second, even
> though they're projected at 48 fps (each frame is projected twice, so the
> flicker frequency is higher).

Well, if we're talking about 35mm (the classic standard), each frame was projected twice (or in some special situations 3 times), but that didn't give an opportunity to meaningfully sync sound other than to a single frame.

> A lot depends on the image size - you're used to a shorter delay when you're
> closer to the person (i.e. their image is larger), but when they're across
> the room you expect a longer delay. 

It's also important to recognize that a large movie theatre presents a problem. Suppose you're in a 500-seat theatre with an 80-foot throw from the projector to the screen. Well, then that's 72 milliseconds (almost 2 frames) for the sound to go from the front of the theatre to the back of the theatre. So, depending on where you sit, you're going to experience it differently, and only one spot will be actually *right*.

In fact, when you thread a 35mm projector, you do so differently depending on the size of the room. There's a loop of film between the optical gate and the sound pickup and you're supposed to vary the size of that loop based on the size of the room. The unit of this adjustment would be the perforation (4 perfs to a frame for normal 35mm motion picture film), so a 10.4 ms adjustment. Trivia note: on 35mm release prints with optical sound, the optical sound is offset 20 frames before the picture, to account for the physical distance in the projector between the sound pickup and the optical gate. 

And for surround sound, the surround channels are delayed from the screen channels such that you hear both sounds at the right time. This is a fixed delay set once in the cinema sound processor (unlike the projector sound loop size that's conceivably variable based on who is threading the projector for every reel of film or movie projected), and of course it's only optimal for a single location in the room.

That said, the human brain will adjust to quite a bit of error, though I vaguely recall that our sensitivity is different for early versus late.

Of course, in the world of digital cinema it's possible to do better than 41.6 ms or 10.4 ms, but that's probably the point of diminishing returns.

--
jhawk at alum.mit.edu
John Hawkinson