AUDIO
SYNC
Any time an electronic signal must be coherent with a
carrier, another signal or the real world, one practice of insuring such
alignment is to use sync signal.
In video, we use a sync signal to insure that the horizontal
scans, the vertical resets, the frequency and phase of the color subcarrier
are all predictable and in phase with other signals with which they must
be mixed. For example, if we are to mix two video signals, the sync reference
on each signal can be processed and aligned so that both signals start
each scan line at the same time.
In modern electronics, the idea of synchronisity is quite
a bit different than in the past. Because of the accuracy of modern crystals
it is possible to keep two independent signals "in sync" for
long periods of time. Modern "syncing" has become more an issue
of establishing and maintaining phase rather than simply frequency.
The double system sync system that we use in the motion
picture business is quite typical of the various ways that many other electrical
devices keep signals in sync in other applications. Originally, both camera
and sound recorder were run off the same 3-phase electrical line. The motors
used in these systems were designed to maintain their RPM based on the
frequency of the AC power source. You can still see, on many stages in
Hollywood, the old 3-Phase power connectors that were used to run the cameras
and full coat recorders. Since both motors would run at exactly the same
RPM (revolutions per minute), when the sound track and picture elements
were mechanically locked together in projection, the actors voices could
easily maintain the exact sync they had when they were recorded. By photographing
and recording the "slapping" of the "sticks", the visual
indication of the sticks closing and the "pop" heard on the sound
track, could be lined up as a reference. If both elements are then started
exactly the same, were on a 3-phase motor system, and were in sync to begin
with, the sound and image would have a perfect marriage.
Then, with the advent of super accurate crystals, it was
possible to run both picture and sound drive motors at exactly the same
speed from day to day, week to week and even month to month. This was only
possible because both the picture stock and the sound recording media were
perforated. If you ran the motor at exactly the same speed from one time
to another, the track and pictures would follow easily because of the interlock
of the drive shaft sprocket and the perforation in the recording or photographing
media.
Full coat recorders were bulky, heavy and were terrible
in the field. As capstan recording became possible, some method needed
to be devised to compensate for the inherent slip in the recording media
as it was fed through the system by the capstan and pinch roller.
The modern (non-timecode) Nagra has a system for effectively
recording sprocket holes in the audio tape as the sound track is being
recorded. This system is done by adding an additional track, usually between
the two sound tracks, and recording a crystal based "pilot" on
this additional track as the recorder is recording the production audio.
Later, when the sound is being duplicated to a "full coat" master,
a system called a "resolver" listens to the pilot tone, compares
it with an internally generated master clock signal, then determines the
exact speed that the capstan motor must run to compensate for any slip.
Typically, this pilot is a 60 hz signal. When the tape
is resolved, if the signal is read at less then 60 hz, the electronics
tells the motor to turn a little faster. If the signal is read at more
than the internal reference of 60 hz, then the electronics tells the motor
to turn a little slower.
It is easy to keep a sound system in sync for many minutes
with a system that has an accuracy of at least two decimal places. For
example, if the original pilot accuracy is 60.002345 hz and the reference
in the resolver is 60.02543, then we will hear noticeable slip in the audio
to image sync within a few minutes. We generally feel that a final frequency
of 60.00XXXX (where XXXX can be almost anything) will be acceptable. With
modern electronics this is a snap.
In certain situations, we can fool a system to run a little
faster or slower by feeding an altered audio sync signal to the recorder
when it is mastering the original sound. This is the case for many motion
picture situations that will be described in the following text.
Timecode
On a time code system, audio sync is maintained exactly
the same way it is on a pilot system. Instead of a 60 hz tone, the additional
track records an FSK (frequency shift keyed) record of a digital time code
signal. This signal contains a running hours:minutes:seconds:frames record
as well as some other information that is called User Bits and a few sync
pulses.
The useful thing about time code is that each 1/2 frame
(usually counted at 30 fps) is uniquely marked on the audio tape in a longitudinal
fashion. In other words, each 1/8 th inch of the tape is uniquely encoded
with a time coded digital word. In video mastered materials, this can be
useful as the video tape can also contain this time code record. We hope
that our film systems will someday be able to easily record time code on
the film stock. Many experiments and prototype systems are being used currently,
but wide spread use of a film recorded timecoded sync system is still in
the future.
Other than the fact that the time code word contains "time
of day and frames" time code information, the speed of a time code
based audio tape is resolved exactly the same way that a pilot recorded
tape is resolved. Instead of resolving the 60 tone, we resolve the 59.94
hz sync bits on each time code word.
Even in a timecode based system, different time based
timecodes can be recorded on the tape. It is important to know what time
base was used so that the tape can be properly resolved.
Digital audio is resolved the same way. Digital data on the tape contains sync bits that are unique and regular. If the player of a digital (DAT or Digital Audio Tape), signal detects the sync bits are being read slower than the sync reference in the player, the tape is made to go faster to compensate.
IN FILM SITUATIONS FOR FILM FINISH
Pilot
A 60 hz tone is recorded on a dedicated section of the
audio tape when it is mastering the audio track. In post production, this
60 hz tone is compared to another 60 hz tone. If there is a difference,
this difference becomes an error signal that alters the capstan speed.
If the film camera is to be run at speed other an exactly
24.00 fps, we can feed a similarly altered pilot tone to the pilot track
on the sound recorder. This is sometimes during video playback and process
projection.
In Video Playback, if it becomes necessary to drive the
camera with a television sync generator, the exact frequency of the camera
speed will be 24.021777 fps. This is due to the fact that some video equipment
will only operate when all of the sync signals are at proper divisions
of the subcarrier. If this is the case, we can manufacture a new pilot
with the same shift.
|
Camera Speed |
Sync or Pilot Frequency |
|
24.00 |
60.00 |
|
24.02 |
60.05 |
|
23.97 |
59.94 |
We usually have devices on board our video systems to
generate the standard pilot shifts that are necessary for creating the
new altered pilot. If not, the post production department can be informed
to resolve the material to an altered reference. If we do our job correctly,
even if the camera is running off speed, the sound will run off speed at
exactly the same rate and the sound and picture will stay in sync.
Timecode and DAT
There is no way to feed a modern timecode Nagra an altered
sync signal during the recording session. If altered sync is necessary,
the only wan to adjust is to adjust in post production. We simply indicate
on the sound log that the camera was running at a different speed. This
speed for our purposes will typically be 24.02 fps.
IN FILM SITUATIONS FOR VIDEO FINISH
When film is being shot for telecine, we have a new set
of problems. Though many shows are photographed at 29.97 fps, many people
prefer to photograph at 24.00 fps. This is usually an artistic decision,
however a show can realize a 20 percent saving in their film budgets if
they stay at 24 fps.
When a film shot at 24.00 fps is transferred to video
on a RANK telecine, the film does not really go though the RANK at 24.00
fps. Because a RANK is operating at 29.97 fps, in order to get all the
information from each frame of film on a frame of video, the 3-2 pull down
system is employed. Basically this means that the film runs though at an
exact ratio of 24 to 30 that really translates to 23.95 to 29.97. Since
no modern camera can run at other than 24.00 or maybe 29.97 or 30.00 and
be at crystal speeds, when the film is transferred, it will be off speed.
The practice of most video finish film shows is the same.
We use a 60.00 hz word time code signal on the audio and run the camera
at 24.00 fps. When the film is run in telecine, it will be a little slower
by the difference between 23.95 and 24.00. Similarly, the sound, time code
or pilot, that was recorded at 60.00 hz will be resolved to a television
sync generator that runs at 59.94. The result is that both sound and picture
are shifted in real time to slightly slower, but since they are shifted
the same, they stay in sync.
It is necessary to use audio sync whenever the video system
may be responsible for variation in the film camera speed from normal operation.
It is necessary for the film camera to maintain .00 fps accuracy to insure
that there will be no slip of audio sync to the camera speed when the audio
is resolved to the film for viewing on a projector.
Use audio sync under the following circumstances.
1. When driving the film camera from a broadcast type
sync generator, Grass Valley or Lenco at 24.02 fps. (When the OPTO-BUFFER
is not in use).
2. When using a REAL-TIME Graphic System that runs at
24.02 fps.
3. When using the Video-Effects System.
4. When using a camera with unknown speed characteristics
such as a visual effects plate camera when connected to an OPTO-BUFFER.
See section on "Unusual Film Camera Synchronization".
5. When performing video playback in a multi-camera show
and it is necessary to drive all the cameras at other than 24.02 fps.
The vertical frequency (frame rate) of 24.021777 Hz causes a slight loss of sync in audio systems running at 24 fps. The error comes to approximately .02 fps, and may not be noticeable over small running times of one to two minutes.
Below is a list of running times and their corresponding
sync loss:
|
Minutes |
1 |
2 |
3 |
4 |
5 |
6 |
| Frame Error |
1.31 |
2.61 |
3.92 |
5.23 |
6.53 |
7.84 |
By adjusting the audio-sync frequency slightly, we compensate for this error.