Re: lip sync

Thread View: comp.multimedia

3 messages

[T] Back to Threads [V] Tree View [G] Group Articles

3 total messages Started by Russ Sasnett Thu, 13 Aug 1992 15:15

#388

Author: Russ Sasnett
Date: Thu, 13 Aug 1992 15:15

47 lines
2660 bytes

speaking as a filmmaker in a former life, i've found
that a professional film editor knows when something
is even 1 frame out of sync (that's at 24 fps for film).

in my experience, a general audience can't tell the 
difference at 1 frame out; they may sense something is wrong
at 2 frames out, but don't know quite what; and at 3 or more
frames out, everybody seems to know what's wrong. 

on a 16mm film projector, the optical sound track is ahead
of the picture by some number of frames; the advance is 20 
for 35mm, i'm not sure about 16mm. this is because the film
must be travelling at a constant velocity when it goes over 
the audio playback head, but the film must be held still for 
picture projection. every 1/48 of second the projector's 
shutter opens and displays a frame frozen in the gate. then 
the shutter closes and the film advances to the next frame 
during the following 1/48 of a second, while the screen is black.
several spring-loaded rollers help to smooth out the jitter 
in the film's velocity before it hits the sound head.

note that there must be exactly 20 frames between the gate and 
the sound pickup to play synchronized sound. when threading, the 
size of the bottom loop is the most critical factor. if the loop
is too small, the sound is early; if the loop is too big, the sound
is late. thanks to armies of amateur operators, whole generations 
of elementary school students have grown up watching 16mm films 
out of sync for this very reason. perhaps this explains a certain
insensitivity to this problem among general audiences ... :)

as to whether early or late makes a difference, i'm told
that being early with sound is much worse than being late;
and logically that would seem to make sense; but in my 
experience it didn't matter. i have known experts who could 
tell pretty quickly whether sound was ahead or behind of 
the pix, but most people don't have a clue about the 
direction of the delay when it's under 1/2 second. i know
i didn't; i always had a tough time syncing up wild tracks 
where there was no slate. back and forth, back and forth 
i'd go ...

try editing film on a Steenbeck flatbed some time. since you
have separate sound and picture tracks, it's real easy to 
slip frames back and forth to experience the effect.

--Russ
rs05@gte.com

Re: lip sync

#408

Author: Rob Koenen (+31 70 332 5310)
Date: Fri, 14 Aug 1992 08:33

28 lines
1868 bytes

In article <1113@ceylon.gte.com>, rs05@gte.com (Russ Sasnett) writes:
> the pix, but most people don't have a clue about the 
> direction of the delay when it's under 1/2 second. i know
> i didn't; i always had a tough time syncing up wild tracks 
> where there was no slate. back and forth, back and forth 
> i'd go ...

We recently performned an informal test to see if people were
able to correctly align audio and video. The tested started with audio
500 msecs ahead of video, or 500 msecs behind. The sequence
contained only a (continuously) talking head. The audio delay
could be changed to get the sound synchronous. Hardly anyone
was able to get it correct within 100 msec, and most scored
between 100 and 200 msec, in either direction.

As several people have noted, tests have shown that audio behind
video is slightly better than having it the other way around.
Such a test could be to have people watch a video with distorted
sound, and see have much they can understand of what is said, while
varying the delay between adio and video. It will be no surprise
that synchronous audio and video turns out to be best.

These tests also show something else: although people may not be 
able to actively synchronise sound and video very well, they do 
give a higher quality rating to material that is synchronous, 
and so they are aware of misalignment.

Rob Koenen

Re: lip sync

#434

Author: Rob Koenen (+31 70 332 5310)
Date: Mon, 17 Aug 1992 10:29

33 lines
1892 bytes

In article <1992Aug14.103330.1@research.ptt.nl>,  I wrote:
> 
> We recently performed an informal test to see if people were
> able to correctly align audio and video. The tested started with audio
> 500 msecs ahead of video, or 500 msecs behind. The sequence
> contained only a (continuously) talking head. The audio delay
> could be changed to get the sound synchronous. Hardly anyone
> was able to get it correct within 100 msec, and most scored
> between 100 and 200 msec, in either direction.

I'm afraid there's something important I forgot to tell about this...

I thought we did the test with original PAL-tapes,
but I now found out that they were coded according
to the CCITT rec. H.261-standard. This is a videophone
standard for coding pictures at around 64 Kbit/sec.
I f you are not familiar with this: it means a lot of
data-reduction. The result is that the frame rate was
around 10 Hz, and the pictures are of less quality than 
PAL. 

I will receive more accurate details in a few days,
but the average error was over 150 msec, in either
direction.

I intend to do the tests again with non-coded material as
soon as I have the equipment back, but this can take a while,
as it's now being used outside of our lab.
I wonder how well people will perform on the original tapes ...

I'm sorry for the inaccuracy in my orignal posting,  

Rob

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads