Tascam HD-P2 and Sony HDR-FX1 success stories?

[Scott Pierce]

Great resource, already, thanks for the forum.

As I'm always on the lookout to improve a double-system sound setup, I
found the HD-P2 to be a great bang for the buck in terms of timecode.
But before I commit to using it in the field, I was wondering if there
was anyone who has taken it onto a narrative film set shooting on an
FX1.

I know the Z1s would be easier, with both free run time code and record
run timecode, as well as XLRs to plug into the sync of the HD-P2, but
we're working with an FX1 this go round.

In the end, we may just go to the standard head and tail slate, but it
would be great to take advantage of the ability to just drop the media
into FCP5 with, if not the media synced, at least the meta data match
to the frame automatically! So if anyone has any practical advise on
the best way to get the FX1 and HD-P2 to play together to their best
potential, I'd appreciate it!

[Steve House]

Originally Posted by Scott Pierce

Great resource, already, thanks for the forum.

As I'm always on the lookout to improve a double-system sound setup, I
found the HD-P2 to be a great bang for the buck in terms of timecode.
But before I commit to using it in the field, I was wondering if there
was anyone who has taken it onto a narrative film set shooting on an
FX1.

I know the Z1s would be easier, with both free run time code and record
run timecode, as well as XLRs to plug into the sync of the HD-P2, but
we're working with an FX1 this go round.

In the end, we may just go to the standard head and tail slate, but it
would be great to take advantage of the ability to just drop the media
into FCP5 with, if not the media synced, at least the meta data match
to the frame automatically! So if anyone has any practical advise on
the best way to get the FX1 and HD-P2 to play together to their best
potential, I'd appreciate it!

While you can set the audio clock to chase video blackburst to keep the clocks in sync and minimize drift, the Tascam doesn't read timecode from the video signal and the FX1 doesn't output timecode directly. So you're still going to have to rely on slates to sync up audio and video by hand in your editor because there won't be frame accurate identical code in the video TC and the audio timestamp.

[Scott Pierce]

Thanks, Steve! Reliable old-school head and tail slates are still the top option on my list.

But leave it to me to not only stick to what works, but keep wondering if there's an even better way. The goal being what I've detailed in the first post above.

I'm getting word that some are using the time pulse coming out of the LANC to run portable sync generators to both the FX1 audio and the HD-P2, but that's only offering either LTC or MIDI, not SMPTE. And that doesn't really give me a good picture on how that improves the workflow FCP without involving conversion software like Gallery or Sebsky Tools.

So keep those cards and letters coming - there should be some field experience out there to prove the solution! Heck, we'll even take theories on this, but be advised that I can't afford to be the guinea pig:)

Thanks again!

[Scott Pierce]

Will probably see the recap so far:

I've gotten a handful of replies in other forums - and from what I'm still trying to piece together, the solutions are rolling in as, each with their own tradeoffs:

1) Old reliable slate clap, align picture to waveforms manually in your NLE
2) Use a timecode slate and jam the signal of the Tascam, align .bwf file data to picture in NLE
3) From LANC, to LTC convertor (such as CLOCKIT SYNC MODULE ACS 401), jam LTC to Tascam. But does the HD-P2 timestamp the .bwf files with LTC? Then when in FCP5, then what?
4) From LANC, to MTC converter (such as Roland SI-80S), convert time pulse to MIDI time code... then would the HD-P2 even read it?
5) Horita generator > synchronize the HD-P2 disconnect and send the output to the FX1's audio input.

The good old-school head and tail slate is still on the top of my list,
and never goes away. We still use it as a holdover from when we worked
less on film and more in digital. Even when it wasn't practically
needed in post, we still clapped a slate, just to give a cue to the
cast and crew that we were working - and it's a Pavlovian thing, that
sound gave them a rhythm that assured that we were making the day.

That said, that's how we've been doing it, and it works - but there
must be a way to automate and take advantage of the .BWF format that
Tascam is giving us and use that metadata to improve our workflows.
Otherwise, if it's not being utilized, what would keep budget-minded
filmmakers from snapping up a Marantz or M-Audio CF unit instead?

So back to the challenge, I guess: how to connect the FX1 to the HD-P2
(either jamming or syncing timecode) to read synchronized timecode.
We'd be using the slate anyway, if not for the reasons above, but also
to safeguard and check against timecode drift on any of the solutions.

I'd really appreciate practical field use for those that have used these two units together, but as I've said, experienced theories - welcome aboard!

In the meantime, it's the slate and a <a
href="http://www.spherico.com/filmtools/LAFCPUG/BWF2XML/index.html">BWF2XML</a>
conversion tool in post. Keeps those cards and letters coming!

[Steve House]

The problem is getting the timecode out of the FX1 to the HDP2. I've heard of devices that will take the timecode from the LANC terminal but have yet to actually find one on the market - if you locate one I'd love to hear about it. IF you can get the code over, the HDP2 will chase external code. So you set the FX1 to use produce Record Run code and the HD will sync up to the incoming code when it comes over from the camera, timestamping the BWF header with a particular from in the take. This "audio chases video" is the reverse of the way code usually works in film.

Psssst - Canon H1 instead of the Sony would make it a no-brainer since it has TC in and out as standard equipment. My dream outfit is an H1 mated to a Sound Devices 702T or 744T. If your budget will sustein it you might want to look at the 702T as it generate TC as will as sync to it and like the 744T it's code module comes from Ambient with all the stability etc of their products.

[Marco Leavitt]

Can't you get a time code generator and feed it into one of the audio tracks of any camcorder and jam it to a 702t? Forgive my ignorance hear, I'm just looking into timecode workflows and so far find it really confusing.

[Steve House]

Originally Posted by Marco Leavitt

Can't you get a time code generator and feed it into one of the audio tracks of any camcorder and jam it to a 702t? Forgive my ignorance hear, I'm just looking into timecode workflows and so far find it really confusing.

The timecode in digital video is, AFIK, Vertical Interval Timecode (VITC) recorded between frames within the video stream itself rather than audio Longitudinal Timecode (LTC) on a parallel track. While you could record timecode signal on one of the audio tracks it would not be encoded into the video clip nor would the NLE know what to do with it during post, since it's just a series of pulses the NLE wouldn't interpret it as timecode so it wouldn't accomplish anything. With file-based digital recorders like the SD702t or 744t the timecode is not recorded linearly on a track alongside the audio like it was on a tape machine in the analog days or on a DAT. Instead the code generator in the recorder runs syncronized to the code generator in the camera and when the recording is started, the code corresponding to the first frame of audio is timestamped into the file header and of course that matches the code stamped on the frame of video that is being recorded at the same instant since the two device's clocks read identically.

The camera and recorder could either be jammed and then allowed to free run or kept connected so the audio sampling clock on the recorder also syncs to the 48kHz sample clock in the camera to even better prevent drift in sync over the duration of the shot. If you're going to free run, which one is the master and which is the slave during the jamming process is kind of a flip of the coin. As I understand from reading in the 702 manual, the norm for DV is audio chases video, taking code out of the camera as the master and slaving the audio recorder to it. My own interpretation is, since the camera's timecode generators may have a tendency to drift a bit when on standby etc, I'd probably vote to jam the recorder to the camera at the start of each take, especially if they're hardwired together and it just takes a button push to do it.

Another possiblity for a camera that doesn't generate code would be to use a Denke timecode slate jammed to the recorder's clock.

NOTE Just a couple hours after posting the above, I found this reference http://www.gallery.co.uk/timetools.html for using LTC on an audio channel with Final Cut Pro!

[Oleg Kaizerman]

Originally Posted by Steve House

The problem is getting the timecode out of the FX1 to the HDP2. I've heard of devices that will take the timecode from the LANC terminal but have yet to actually find one on the market - if you locate one I'd love to hear about it. IF you can get the code over, the HDP2 will chase external code. So you set the FX1 to use produce Record Run code and the HD will sync up to the incoming code when it comes over from the camera, timestamping the BWF header with a particular from in the take. This "audio chases video" is the reverse of the way code usually works in film.

Psssst - Canon H1 instead of the Sony would make it a no-brainer since it has TC in and out as standard equipment. My dream outfit is an H1 mated to a Sound Devices 702T or 744T. If your budget will sustein it you might want to look at the 702T as it generate TC as will as sync to it and like the 744T it's code module comes from Ambient with all the stability etc of their products.

check with ambient .de
they show one

[A. J. deLange]

I have a cable from a company in the UK, AVIT Research, which converts the LANC time code to MIDI. This can be plugged into a MIDI to SMPTE converter (e.g. MOTU Timepiece) which then will sync the HD-P2 (which can simultaneously be clocked from the camera's video).

The pertinent questions are: But does the HD-P2 timestamp the .bwf files with LTC? Then when in FCP5, then what?

The answer to the first is "Yes" (as demonstrated by the fact that software supplied with the recorder will "conform" a batch of non contiguous BWFs into a single one with each cut in it's proper place in time). The answer to the second is "I don't know!" There is a utility called BWF2XML which is supposed to convert the BWF files to XML files that FCP can read but I can't make it work. It keeps responding with "BWF file parsing error - DF variable not set" or something very like this. This suggests that the problem has to do with the software not knowing whether the TC is to be generated in DF or NDF mode though that's just a guess. I'm going to try contacting the developer but if anyone else knows how to do this I'd love to hear about it.

[Edited to add name of converter manufacturer]

[Steve House]

Originally Posted by A. J. deLange

I have a cable from a company in the UK, AVIT Research, which converts the LANC time code to MIDI. This can be plugged into a MIDI to SMPTE converter (e.g. MOTU Timepiece) which then will sync the HD-P2 (which can simultaneously be clocked from the camera's video).

Are you referring to the MOTU Midi Timepiece AV or the MOTU Digital Timepiece? If the latter, doesn't the Digital Timepiece accept standard NTSC as the master clock input?

I think one source of confusion - certainly for me - is there's sync and then there's sync and it's often unclear what sort of sync we're talking about. There's sample clock sync where the audio recorder's sample clock is slaved the the video camera's so the two devices run at exactly the same speed and then there's timecode sync where a frame of audio is timestamped with a set of numbers that exactly match the numbers stamping the matching frame of video. Sample clock sync would be like having two clocks on the wall locked together to run at exactly the same speed. Timecode sync would be setting them to read exactly the same time of day.

A couple of questions unclear in my mind. If a typical prosumer video camera such as an XL2 or Sony Z1 or FX1 is generating timecode, does its analog video output when monitoring during recording carry the timecode or not? If I send that video to a MOTU Digital Timepiece, does it extract the actual timecode numbers or just the clock rate? If the video is sent to the TASCAM recorder, again does the recorder read the timecode numbers and jam its own TC generator to the same values or is it just slaving its sample clock to the video sample rate?

In a typical double system sound scenario it seems we have two separate issues to contend with. We need to jam the timecode values in the camera and the audio recorder to read exactly the same numbers at any given point in time and we also need to sync their sample clocks so they run at exactly the same speed. Given a camera such as the XL2 that does not have dedicated TC input and output ports and a recorder such as the Tascam or a Sound Devices 744T or 702T, how can we best achieve both goals?

[A. J. deLange]

Steve,

Doesn't sound to me as if you are confused. The clock analogy is a good one.

I was referring to the Digital Timepiece. It will resolve its sample clock to SMPTE time code or to video if video is available. In the former case the derived sample clock won't be quite as stable as in the latter and is referred to as 'near sample accurate'. In the latter greater accuracy is acheived but there is still some potential for a bit of "jitter". This would be at a low enough level as to be imperceptible. In other words, the clock in the recorder is now running at the same average rate as the clock in the camera. Now to get them set to the same time. This is what the time code does but the way it does it is not by recording the time code signal (which you can, of course, do on a spare audio channel if you wish) but rather by time stamping the beginning of the audio file with the time at which the file begins according to the incoming timecode. If you can get Quicktime to read the timecode stamp (and other meta data concerning sample rate and time code format) it can then generate a time code track for the clip and then FCP can use the timecode from the timecode track. It can do this because knowing the sample rate of the audio and that it is tied to the frame rate of the video it can calculate the exact sample numbers at which the video frames start simply by counting the audio samples.

The problem with the Tascam HD-P2 is that while it advertises that it is recording BWF files the 'bext' chunk (which contains the "broacast extension" meta data) is not in the expected format and BWF2XML, the only program for adding the timecode track and putting the files in a QT wrapper I'm aware of, is unable to read the timestamp. I'm still investigating this.

Do the prosumer cameras put VITC on their analog outputs? I doubt it but it could depend on the camera. The MOTU Timepiece is capable of detecting VITC but does not lock up to video from the XL2, for example, in this mode.

With the TASCAM unit you have a choice. It can be set to follow the timecode and generate sample clock from it or clock by the video but have the time code generator follow incoming timecode or jam to a particular preset (by the user) timecode when it is received and freewheel indefinitiely thereafter.

The XL2's video is the signal necessary to lock sample clocks i.e. to get the rates the same. The problem is in getting the timecode out of the camera. It's there on the LANC port but needs to be converted to SMPTE for units like the TASCAM HD-P2, the MOTU Traveller or any of the other recording devices that take SMPTE LTC timecode. The conversion is simple and there used to be a little box made which did exactly this but it is no longer made. The only way I know to do it now is via LANC to MIDI and then MIDI to SMPTE conversion.

[Steve House]

AJ - wasn't thinking about how to get a Tascam or SD recorder to actually record timecode with the audio stream but rather the jamming process required to set the clocks together - if the camera's timecode generator currently says it's 00:00:32:37;15, how do we get the recorder's own TC generator to set itself to also read 00:00:32:37;15 at the same precise instant? With a camera like the Canon H1 that has TC out it's a no brainer - connect a cable from the camera TC out to the recorder TC in and with the SD recorders at least hit the "jam" button and presto! - clocks sync'ed. SD recommends setting the camera to generate record-run code, rolling camera first, and then before rolling audio jamming the audio recorder to incoming camera TC once the camera has speed. But how to do it on a camera without a dedicated TC output?

[Oleg Kaizerman]

regullar slate , free run , small external tc generator for the tascam that can take external tc or no generator for sd .working every time for me .
you slate about ones in 2-3 hours and later automatically sync the all shotes on nle .
for cameras like fx-1 which not able run in free run you have to supply tc to audio channel then later replace the existing tc whith the audible one on nle ( working with avid) or pull out lank tc and contentiously supply it to your recorder

[A. J. deLange]

I guess I wasn't too clear on that. All that is necessary to do is pass timecode to the TASCAM with timecode enabled and place it it timecode chase mode. If valid timecode is coming in the TASCAM will follow it. If you press the record button the recording will be tagged with the starting time code (in a strange format, however: the number of samples since 00:00:00:00 but this is what the BWF spec requires). If no valid time code is coming in and you press the record button, nothing happens until time code starts to count up (i.e. you press record on an XL-H1) at which time the TASCAM locks and recording begins. I.e. you can start the audio recorder just by starting the video camera. Unfortunately it doesn't work the other way as the same time code is transmitted over and over when an XL-H1 stops and the TASCAM goes into freewheel (however you set the freewheel control) and keeps recording until you press "stop". The owner's manual for the HD-P2 is on the web at http://www.tascam.com/Products/hdp2.html though the section on timecode is a little sketchy.

[Steve House]

Originally Posted by A. J. deLange

I guess I wasn't too clear on that. All that is necessary to do is pass timecode to the TASCAM with timecode enabled and place it it timecode chase mode. If valid timecode is coming in the TASCAM will follow it. If you press the record button the recording will be tagged with the starting time code (in a strange format, however: the number of samples since 00:00:00:00 but this is what the BWF spec requires). If no valid time code is coming in and you press the record button, nothing happens until time code starts to count up (i.e. you press record on an XL-H1) at which time the TASCAM locks and recording begins. I.e. you can start the audio recorder just by starting the video camera. Unfortunately it doesn't work the other way as the same time code is transmitted over and over when an XL-H1 stops and the TASCAM goes into freewheel (however you set the freewheel control) and keeps recording until you press "stop". The owner's manual for the HD-P2 is on the web at http://www.tascam.com/Products/hdp2.html though the section on timecode is a little sketchy.

Ahhh - but how can you recover the code from a camera such as an XL2 that does not have a dedicated TC output to use in a similar manner? I checked the company you mentioned earlier for the LANC cable and it looks like you need to "roll your own" with their SDK to use it. They even go so far as to say they won't sell the cable to anyone who isn't a registered purchaser of the SDK. I found a box that works with it to output LTC but it's really pricy!

Haven't purchased either one yet but have been considering an XL2 and a Sound Devices 702T (same timecode functions as the 744T) but haven't figured out that bit of the puzzle. Almost at the point of thinking the film style way is best for cameras without TC out, taking a cue from practices common with film cameras that didn't print timecode to the filmstock, and use the audio recorder as the master jamming to a Denke slate, slating each take, dropping the audio into the timeline at its timestamp and then lining up picture to it by aligning the visual of the slate's displayed code to the matching point in the timeline. A Denke is pricey but the only box I've found that takes LANC from the camera and splits out the timecode as LTC costs about the same amount (and is no where near as sexy on the set <grin>).