Bizarro world sync issue - Please please help!

Main Page | Report this Page

Movies Forum Index » Movie Production (Sound) Forum » Bizarro world sync issue - Please please help!

Page 1 of 1

Author

Message

Anig Browl

Posted: Wed Jul 07, 2004 12:20 am

Guest

OK, all you pro dudes, I need to beg your assistance for a horrible
sync problem I am having. We are supposed have a rough cut on Thursday
(yes, this thursday) and finish the film by around the 25th. Right now
I am slowing things down and I don't like it. Let me tell you the
whole depressing story.

1. It's a short feature, 25 minutes, 35mm. We recorded sound on DAT
with 2 production microhpnes at all times, and in the main it has come
out extremely sweet (yay me). 48.0 KHz sample rate. But, there was one
mistake on the production - I had the DAT timecode set at 24 fps the
whole time. However, this may not be relevant at all, because...

2. We did a telecine transfer at the San Francisco Academy of Art,
where the director is a grad student. The resulting video was then
transferred from mini-DV to Avid at 29.97. I, meanwhile, was doing the
DAT transfer, from a Sony PCM-R300 into Pro Tools. The PCM-R300 does
not output timecode at all, nor does it have any pulkl up/down options
- it just played everything out at 48KHz. Mistake #2: I set the pro
tools projects to be 24 FPS, though not in all cases before recording
the audio. I forget: I've been working 12-15 hour days for 3 weeks.
Mistake #3: I honestly can't remember if Pro Tools was syncing to the
DAT S/PDIF signal during recording - I think so though - or to
'internal', the output of the nearby MOTU MIDI timepiece AV (normally
48.0k/30 fps, no pull up or down). didn't notice any pitch warbling
during playback.

(Now, as far as I can tell, I could have set Pro Tools to be 117 1/2
frames per second during the recording part and it wouldn't have given
a hoot since it was not getting any incoming timecode. Everything
sounds just fine coming off the tape.)

3. Flushed with enthusiasm, I marked and subclipped all the scenes and
then exported them all as AIFF files keyed to scene/shot/take. The
editor looooves me. Until we go to import them to Avid, and find a lot
of them don't sync. My major major mistake: while I was doing this I
was not aware that the MIDI Time piece AV is providing timecode sync
to Pro Tools. Foolishly, I assumed that because we were originating
with film and going back to it, I could just set everything to 24fps
and be fine (you can hit me now). More foolishly, I assumed that the
Digidesign 888 i/o box had been following whatever frame rate I was
setting on the screen in PT, instead of slaving to an external box
(possibly because said box was switched off when I was first let into
the editing room).

Now, it may well be that these machines are at the aformentioned
Academy of Art - meaning that other people are in using the Pro Tools
room between my sessions. Naturally, the first move we made was to
consult with the lab techs and then the instructors. After explaining
all the above to them, they have no idea what the exact problem is
(reassuring, eh?).

Here's my current situation: it seems that everything I took from DATs
3 & 4 (of 4) is fine and syncs perfectly - the editor is going to
check shot by shot tomorrow but a random spot check of 5 scenes from
each tape suggests that everything is OK, even though these were
recorded and exported with Pro Tools set to 24 FPS. We are not sure
why ( and I hate that), but from a practical standpoint, that audio is
OK and we are going to keep using it until we find a reason not to. I
have to apply noise reduction and all that good stuff soon, so I hope
I am OK committing to those recordings.

Stuff that came from DATS 1 & 2 is waaaay out of sync, it drifts off
by (something like) 5 seconds out of 60? during long shots. The audio
is running faster than the picture. This is a lot, no? Meanwhile, we
have tried re-exporting at 29.97 drop and non-drop, 30 fps, 24, etc.
etc. and we just cannot get the original audio to sync up with a
quicktime test scene we have (whereas another scene shot later in the
prodcution on one of the other tapes syncs up immediately with another
quicktime test video).

The scary bit: I took the test scene I had and re-recorded that audio
from the dAT today, again trying as many combinations as I could find
of recording - at 30 and dropping out at 29.97 pull down (which I
should have done from the outset), at 29.97 and dropping out as 20, at
30 and dropping out at 24 and so on and on and on. Nothing seems to
fit. The audio mentioned above that does fit sounds fine and normal;
when I futz about with recording at one timecode and
exporting/bouncing at another then the pitch changes around in the
proportions you'd expect, eg everyone sounds slightly older and larger
when 30 is reduced to 29.97 non-drop (and it still doesn't sync).

I am at my wits' end and while I accept that I might as well just
re-transfer those two DATs, I'm starting to get really worried about
whether I am ever going to find the 'magic' combination which worked
on the other tapes and is syncing up fine. And as well, nobody seems
to have a clear answer on how to deal with this when we transfer back
to film. The telecine video doesn't have timecode or keycode because
it's all going to be telecined again on a better machine so the
director didn't bother (and now I wish he had, or that I had known to
ask him).

Well, I hate 'magic' combinations of settings. I'm an engineer and it
makes me feel stupid not to know what's going on, not to mention that
I hate not having an answer for the director. I feel terrible even
though I'm working for free and have done (I'm told) an 'outstanding'
job on the production recordings. He's relaxed about it and is not
giving me a hard time, but own sense of perfectionism is.

Have any of you Hollywood sound mavens run into this?? You now have
all the information I have (really all - I used to be a computer tech,
so I've been as methodical as I can about tackling it). I just don't
understand what Pro Tools is up to, I'm exhausted physically, the
results seem to defy logic, and worst of all I really care about this
movie and I'm terribly proud of the recordings I got, so seeing (some
of) them not sync up with the picture is painful.

Please help <:-(

Thanks,

eddy robinson
handmade sound <-- and feeling a bit too handmade today!

Jay Rose CAS

Posted: Wed Jul 07, 2004 9:20 am

Guest

Quote:

Stuff that came from DATS 1 & 2 is waaaay out of sync, it drifts off
by (something like) 5 seconds out of 60? during long shots. The audio
is running faster than the picture.

That's your key. The drift is too much to be a pull-up/down issue. If it's
consistent at that ratio, it's either sample rate (playing a 44.1 kHz DAT
at 48 kHz) or 24 fps vs 29.97/30 fps timecode. IMHO the former is much
more likely - it can be something as simple as giving the playback
hardware the wrong sample rate setting or word clock - particularly if
your playback DAT deck didn't handle timecode at all. But I'm not a
ProTools guru and don't know how inflexible it is about timecode
settings*.

(*Hey, you PT gurus: once you set up a session, can you change timecode
settings without messing up sync? My DAW uses an internal sample-based
timeline so that's no problem - it calculates TC from it on the fly, based
on whatever counting system you want - but I know DP and Logic will lose
sync between regions if you mess with TC settings...)

--
Correct address is spell out the letter j, AT dplaydahtcom
Clio- and Emmy-winning sound design
Learn audio for video at www.dplay.com

Philip Perkins

Posted: Wed Jul 07, 2004 10:18 am

Guest

You have been very industrious in a science-project sort of way as you
search for a magic bullet to correct your original mistakes. Since
this is a student film you are working on, chalk it off to a learning
experience about the value of doing things right the
first time, and go do what you have to do. This means getting your
DAT audio to play back at the correct speed, probably by doing a
pull-down transfer. This seems to be the best idea since half your
audio is working sync wise, so you don't want to change the sync set
up of your ProTools. It will be some tedious work but should come
out right in the end. One other thing to consider--esp. if your
properly pulled-down retransfer doesn't work out, is that there was a
CAMERA speed issue as well on some rolls. This has turned out to be
the problem several times for me over the years. There are solutions
to this too but they are much more complex. You've just given
yourself a pretty thorough crash-course in how sync does NOT work in
audio post--now take a deep breath,
make a carefully considered to-do list and go make your movie.

good luck

Philip Perkins CAS

John Gilman

Posted: Wed Jul 07, 2004 11:38 am

Guest

On 6 Jul 2004 23:20:08 -0700, anigbrowl@softhome.net (Anig Browl) wrote:

<snip>

Quote:

Not getting into any of the other issues, but 5 seconds out of 60 is
amazingly close to the difference between 44.1 and 48. Is it possible that
Dats 1 and 2 were either inadvertently transferred at the wrong clock rate
or recorded wrong. If the audio is running faster than the picture, then my
guess is that they were recorded at 44.1 and transferred at 48.

Or not.

--
John Gilman
zoundz@xxxReMoVexxxnewscene.com

Brad Harper

Posted: Wed Jul 07, 2004 4:50 pm

Guest

If the audio sounds correct on the tapes it is possible that the takes are
mis-slated and they are trying to match take 5 sound to take 6 picture. I
had this happen at a very reputable transfer house and after they called me
in to show me how my sound was drifting I discovered that on a camera roll
change the take number on the slate went from 3 to 5. Due to the fact that
this happened at a roll change they didn't catch it and decided to blame me.

The other possibility is that the camera could have been running at a
different frame rate.

Brad Harper

Quote:

Stuff that came from DATS 1 & 2 is waaaay out of sync, it drifts off
by (something like) 5 seconds out of 60? during long shots. The audio
is running faster than the picture. This is a lot, no? Meanwhile, we
have tried re-exporting at 29.97 drop and non-drop, 30 fps, 24, etc.
etc. and we just cannot get the original audio to sync up with a
quicktime test scene we have (whereas another scene shot later in the
prodcution on one of the other tapes syncs up immediately with another
quicktime test video).

The scary bit: I took the test scene I had and re-recorded that audio
from the dAT today, again trying as many combinations as I could find
of recording - at 30 and dropping out at 29.97 pull down (which I
should have done from the outset), at 29.97 and dropping out as 20, at
30 and dropping out at 24 and so on and on and on. Nothing seems to
fit. The audio mentioned above that does fit sounds fine and normal;
when I futz about with recording at one timecode and
exporting/bouncing at another then the pitch changes around in the
proportions you'd expect, eg everyone sounds slightly older and larger
when 30 is reduced to 29.97 non-drop (and it still doesn't sync).

I am at my wits' end and while I accept that I might as well just
re-transfer those two DATs, I'm starting to get really worried about
whether I am ever going to find the 'magic' combination which worked
on the other tapes and is syncing up fine. And as well, nobody seems
to have a clear answer on how to deal with this when we transfer back
to film. The telecine video doesn't have timecode or keycode because
it's all going to be telecined again on a better machine so the
director didn't bother (and now I wish he had, or that I had known to
ask him).

Well, I hate 'magic' combinations of settings. I'm an engineer and it
makes me feel stupid not to know what's going on, not to mention that
I hate not having an answer for the director. I feel terrible even
though I'm working for free and have done (I'm told) an 'outstanding'
job on the production recordings. He's relaxed about it and is not
giving me a hard time, but own sense of perfectionism is.

Have any of you Hollywood sound mavens run into this?? You now have
all the information I have (really all - I used to be a computer tech,
so I've been as methodical as I can about tackling it). I just don't
understand what Pro Tools is up to, I'm exhausted physically, the
results seem to defy logic, and worst of all I really care about this
movie and I'm terribly proud of the recordings I got, so seeing (some
of) them not sync up with the picture is painful.

Please help <:-(

Thanks,

eddy robinson
handmade sound <-- and feeling a bit too handmade today!

Anig Browl

Posted: Wed Jul 07, 2004 8:34 pm

Guest

Wow, thanks for all the helpful replies. You guys rock. Well, we are
still having the problem, so I'll quote relevant bits from all the
replies below...

Jay Rose wrote:

Quote:

The drift is too much to be a pull-up/down issue. If it's
consistent at that ratio, it's either sample rate (playing a 44.1 kHz

DAT
at 48 kHz) or 24 fps vs 29.97/30 fps timecode. IMHO the former is much
more likely - it can be something as simple as giving the playback
hardware the wrong sample rate setting or word clock - particularly if
your playback DAT deck didn't handle timecode at all.

You'll be happy to know I consulted your Sound for DV book extensively
since getting into this mess, so your advice has not gone unheeded!

Well, I really don't think it's 44.1 vs 48. I'm pretty anal about
checking my gear between shots and not letting anyone else handle it
during shooting. For a start the sound would be almost 10% faster and
that would be very noticeable. Plus I had my eyes glued to the DAT
output and it was on 48k (not 48.048 or anything else) all the time. I
recorded another DAT the other day with mixed sample-rate stuff on it
and the display on my playback DAT switches automatically, so I feel
confident about that. The playback DAT does not have a word clock
input, when recording the normal procedure is to sync Pro Tools to the
48KHz digital signal itself @ 30FPS.

We did that again this morning and re-recorded the 'bad' scenes, and
as I expected when we switched to 29.97 & 48 KHz pull down sample
rate, the bounce in PT was noticeably pitched down so the actors'
voices all sounded slightly deeper, as you'd expect. Nonetheless this
still doesn't sync up. I even tried the experiment of exporting the
should-be-correct audio from Pro Tools at 48048 and then telling Avid
to treat it as straight 48k (ie play it back a little slowly), but no
joy. Shit.

"Brad Harper" <bradharper@comcast.net> wrote...

Quote:

If the audio sounds correct on the tapes it is possible that the takes are
mis-slated and they are trying to match take 5 sound to take 6 picture.

Maaaaaybe, but I kept notes on every take and we always called slate -
although it was (mainly) a student crew, I think this was only
reflected in the lack of pay for the work, it was a very tight
operation from start to wrap. Ditto on the camera speed, it could
conceivably have been 25 FPS on some shots but I'd be very surprised
if the DP had overlooked that, he has commercial and feature
experience. Also the director watched the whole telecine process so I
imagine he'd have spotted a discrepancy.

Philip Perkins writes...

Quote:

You have been very industrious in a science-project sort of way as you
search for a magic bullet to correct your original mistakes.

Hehehe :-)

Quote:

Since this is a student film you are working on, chalk it off to a learning
experience about the value of doing things right the first time, and go do
what you have to do. This means getting your DAT audio to play back at the
correct speed, probably by doing a pull-down transfer.

Oh indeed - I care about getting this right, rather than finding a
quick fix. So this morning I did indeed retransfer the stuff as
described above, but as you see it still doesn't work. However....

Quote:

This seems to be the best idea since half your audio is working sync wise, so
you don't want to change the sync set up of your ProTools.

Ech, it's not mine - I hate PT. It's ugly, baroque, and I think the
only reason it performs so well is that they've thrown so much money
at the problem. I used to work for a small synthesizer company and I
can tell you that what they do with the DSP isn't that impressive.

Quote:

One other thing to consider--esp. if your properly pulled-down retransfer doesn't work out, is that there was a CAMERA speed issue as well on some rolls. This has turned out to be the problem several times for me over the years.

This just *might* be the case. We used a 1000 foot magazine on about
40-45% of the shots, vs. a 400 foot one for the rest, and apparently
that large magazine takes a few feet to get up to...camera speed. And
you know, that would account for some of the oddness that the editor
and I noticed on the first day, like the sync falling radically behind
in the first 4-5 seconds and then not getting much worse as the take
went on. We're off to compare shot lists now, and I'm hoping that if
they match up we can lip sync instead of syncing to clapper.

Goddamn, if it turns out to be a camera issue I think I'll just spit!

Audiomixer

Posted: Wed Jul 07, 2004 9:39 pm

Guest

I don't know if this helps, but an issue I've had before is the actual
sample rate and the flag where different . You can have a 48K flagged dat
that is actually recorded 44.1 and vice versa. The only way I have been able
to get around it is do an analog transfer at the proper sample rate.

Philip Perkins

Posted: Thu Jul 08, 2004 1:01 am

Guest

Quote:

Goddamn, if it turns out to be a camera issue I think I'll just spit!

Well get ready to pucker up, dude! If a motion picture camera with
any length
load takes longer than 3 or 4 frames to lock @ sound speed then it is
either very old,
very poorly maintained or running on a very low battery. There are a
few
ways that the camera can be made to (inadvertently) run off speed, but
unless
you can determine what the real frame rate was, and then design a
playback
system to compensate for this (either in a computer or via
speed-shifting a
playback deck) then you still have to get the movie made with what you
have.
The "El Mariachi" fix comes to mind, where you just have to keep
pulling
up and resyncing every so often in the middle of shots (I hope your
film has lots of cuts).
I've had to do this due to bad transfers and no time for either
research or re-transfer.
It sucked but we got through it. I hope you find your magic bullet.

Philip Perkins CAS

Courtney Goodin

Posted: Thu Jul 08, 2004 11:14 am

Guest

"Anig Browl" <anigbrowl@softhome.net> wrote in message
news:3791530f.0407071834.487cdaf7@posting.google.com...

Quote:

We did that again this morning and re-recorded the 'bad' scenes, and
as I expected when we switched to 29.97 & 48 KHz pull down sample
rate, the bounce in PT was noticeably pitched down so the actors'
voices all sounded slightly deeper, as you'd expect. Nonetheless this
still doesn't sync up.

One thing that seems strange in your statement, You should not be able to
hear a difference between 30.00 and 29.97 or a .1% pull down. Voices should
not sound noticeably deeper. Most humans can't perceive a pitch difference
of 0.1%. If you had a constant tone and an electronic tuning meter you
might be able to tell the difference on the meter, but the difference in
pitch should not be perceptible to the human ear.. Perhaps some setting in
Pro-Tools or the Sync Reference was set wrong and the software was pulling
the sound down too far. Much more than .1%.

Most non TC- DAT machines will play back 48khz Dat tapes pretty accurately.
Just play it back wild, with no sync reference and bring it into an audio
editor. Then try syncing it up straight, then slowing it down 0.1% and see
how it holds sync throughout a long shot. Just sync the slates to the clap
stick and ignore the Time Code. If neither the straight or pulled down 0.1%
stay in sync, it was probably a Camera problem. There is no easy way to
tell if the camera was running off speed unless florescent lights start to
flicker or strobe in the scene or the exposure fluctuates visibly. If it is
just running consistently off speed and no florescent or 60 Hz HMI lights
are in the scene you probably won't be able to tell.

---Courtney Goodin

Anig Browl

Posted: Thu Jul 08, 2004 11:31 am

Guest

spamiser@yahoo.com (Philip Perkins) wrote:

Quote:

Well, we did not sync camera and production sound, but I think you
meant to lock to 24 fps? But yes, there were camera battery problems
on the set, the monitor image would get more and more blurred. Also,
we used both 1000' and 400' magazines, and the shots that were done on
the 1000' rolls correlated very closely with the ones which have sync
problems (some audio files I can't recall if there were or not).

The only problem with the slow camera speed theory is that the audio
is running faster than picture, when you'd think at playback the
picture would be speeded up and running ahead of the audio -
suggesting that the film was in fact running slightly fast? I don't
know how the Arri 3l works (or much about cameras in general), perhaps
a governor that holds the motor back was not perfectly engaged.
Anyhow, the director doesn't think it's even my fault at this point...

Quote:

The "El Mariachi" fix comes to mind, where you just have to keep
pulling up and resyncing every so often in the middle of shots (I hope your
film has lots of cuts).

Yeah. But of course it had to happen on the longest scenes with the
most dialog Smile

And that is exactly what I am going to have to do. You
will be amused to know that we have until about the 25th to finish
this film, and that the footage that made the cut is being
re-telecined next week in LA, so basically I have 14 days to do the
sync, sound design and mixing and I need a computer upgrade too.

Good thing my production recordings came out so well - although it's
25 minutes with ~200 shots and lots of dialog (think asian teen
sci-fi) it seems like we may not have to do any ADR at all. Although I
wish I had taped a few more rehearsals for a few scenes where camera
noise was a problem. Well, you live and learn.

Thanks for all the advice, I *really* appreciate it.

Anig Browl

Posted: Thu Jul 08, 2004 9:05 pm

Guest

"Courtney Goodin" <cmgoodin@pacbell.net> wrote in message news:<vhfHc.7944$t6.938@newssvr25.news.prodigy.com>...

Quote:

Well bear in mind that for the sake of experiment we did some with
both 29.97 and pull down engaged. I agree about most people not being
able to hear the difference, but it doesn't seem that strange to me;
perhaps it comes from being a music engineer before and having had to
deal with people going out of tune.

Out of curiosity, I've just done some deliberate pitch stretches by
that factor and I believe I can hear the difference...but I'd have to
have some one A/B test me to be 100% sure, perhaps I'll have time
after the project is over. We've decided to give up further analysis
for now and just get on with it. Which means I have 13 days to produce
a soundtrack out of this horrible mess. Oh well, at least I have a new
computer tomorrow.

Despite the fact that I'm not making any money (I decided to spend the
year 'earning my stripes' in the film environment and investing in
gear), I'm starting to wonder if I should find some naive eager beaver
to intern for me. I can't take on any more work between now and the
end of the year, I still have to work on an episode-lenght DV project
and then a feature film in Septermber, all while holding down a
(fortunately very flexible) day job. Yikes.

Page 1 of 1 All times are GMT - 5 Hours The time now is Sat Nov 22, 2008 10:50 am Movies Forum Index