What is dual-system audio?
Dual-system audio is one name for the technique of recording audio and video on separate devices, and then combining them in post-production. Since the quality of built-in mics and audio systems of cameras are generally not nearly as good as the video, dual-system is a convenient way to get high-quality sound. It is particularly effective for video coming from a DSLR like the wildly popular Canon 5D Mark II.
The price to pay for these benefits is a small one: you need to synchronize the audio and video in post-production. This can be done manually or automatically using our PluralEyes product.
The problem with Final Cut Pro
Many non-traditional cameras shoot video at 30 fps instead of the more usual (in some parts of the world) NTSC rate of 29.97 frames per second (fps). If you are using Final Cut Pro with such video there is one quirk that you need to be aware of. This has been discussed in recent blog postings (see, for example the comment at 9:45 of the Zacuto video here). The quirk has come to be known as the 99.9% problem: when audio is combined with 30 fps video, the audio speed (sometimes) needs to be adjusted by the NTSC factor of 99.9% (actually 1000/1001).
Fortunately there is a simple fix for this: use an Easy Setup for 30 fps. Details below, but first some background.
Background
Video from a DSLR and other still cameras is typically recorded with a frame rate of 30 fps. If you do a lot of work with NTSC video, your default settings in Final Cut might be something like HDV 1080p30 which has a framerate of 29.97 fps. When you work with 30 fps material, you might expect that ensuring that the sequence frame rate is 30 (and not 29.97) would be enough to make things work. But what happens is that when you add an audio track to the timeline, Final Cut retains some memory of the NTSC settings and applies an NTSC adjustment to the playback speed of the audio. This seems like a bug to me--I can't imagine any situation where I would want it to do this. Fortunately you can prevent this from happening by using the right kind of Easy Setup.
The pre-packaged, easy way
If you are editing video from a Canon 5D Mark II using the ProRes 422 codec, you can download the Easy Setup here and copy it to /Library/Application Support/Final Cut Pro System Support/Custom Settings. Choose this Easy Setup, restart Final Cut and all should be good.
If you are using a different kind of video that is using a non-NTSC framerate, the following takes you through the steps to create the Easy Setup. You'll need to make changes that are appropriate for your video format, but they should be obvious.
The do it yourself way
- Create a suitable Easy Setup
- Select that Easy Setup
- Re-start Final Cut Pro
Here are the details on how to create an Easy Setup for video from a Canon 5D Mark II that has been transcoded to ProRes 422 (which seems to be the consensus recommendation for how to edit 5DMKII video, at least, if you've got FCP 6). We're creating a new Easy Setup because none of the ones that ships with Final Cut is suitable. The following video tutorial shows both the problem and how to fix it. [Note: In the tutorial I suggest changing the audio sample rate to 44.1 kHz. It makes more sense to use the sample rate of your externally recorded audio, which is probably 48 kHz.]
Go to the menu item Final Cut Pro > Audio/Video Settings... > Sequence Presets. Pick Apple ProRes 422 30p 48kHz, because it's close to what we want to end up with, and choose Duplicate...
Type in something appropriate for the name and description as shown and make two changes. The important one is to change the Editing Timebase to 30 fps. Set the Audio Settings > Rate to the rate of your externally recorded audio (probably 48 kHz).
Choose OK and then go to the Summary tab and choose Create Easy Setup.
Type in an appropriate description,
and select this Easy Setup.
Restart Final Cut Pro. For any newly created sequences, the audio and video will play at the same rate with no need for a 99.9% speed change.
Posts
16 comments:
Great info. Thanks!
Bruce, Thank you for this! It fixes the 5d's real Achilles heel
Great piece. I think most folks would prefer to use 48K audio over 44.1, however. If you're going to all the trouble of dual system recording, why lower your audio quality? Is there something about how this affects the frame rate? All my post work is done at 48K.
Thanks!
Good point. I used 44.1 kHz only because that is the format that comes from the 5D. It should not have any effect on the frame issues that the post was about, and it would make sense to set the audio rate to match that of the dual-system audio.
Thanks Bruce!
We are still having some troubles after testing your solution with the new Final Cut (FCS 3). Probably we are doing something wrong. For this particular project we are editing in 720p. We converted our Mark II video to 720p Apple ProRes (using mpegstreamclip) and then imported all prores-video into a new FC sequence together with our separate audio track (recorded on a Olympus LS-10 at 44.1khz). After trying your solution we are still having some audio drifting. If we speed the audio up 105% it gets back in sync again. Could it be the new version of Final Cut, or what could be the problem?
/Ulf Textorius - Linkin Media
@Ulf: The likely cause is that the LS-10 clock is not very accurate. We haven't tested the LS-10 specifically but we have found that other audio recorders can have significant drift because they run a little fast or slow.
We're working on a solution for this, but in the meantime all I can suggest is that you use the Final Cut speed change command to modify the speed.
If you do a careful recording of the LS-10 against the 5D for as long a clip as you can, with hand claps at the ends to find the sync, and determine the speed adjustment factor from that you can probably use the same adjustment every time.
Aha! Very good info. Thanks again Bruce. I'll try to determine the exact speed adjustment needed right away!
I couldn't get this to work. I downloaded the easy set up and went through the process of making my own. Neither worked for me.
I'm pretty sure I'm doing everything as described. Is there something else I should keep in mind?
@Anonymous: Can you be more specific about the way it didn't work? What results did you get? Also, what camera and audio recorder are you using?
Sorry but how is this different than just changing the speed of the audio to 99.9%?
Why change the speed if you don't have to? And changing the speed doesn't always work: it depends what kind of project you were working on previously. So you have to pay attention to what your settings are anyway. Seems like doing it the right way in the first place is no harder and is more reliable than hoping that two wrongs will make a right.
This is great. One question, I also need to add footage and sound that I shot with my EX1 with my 5D and external sound. How will this effect that footage? Or will it? Thanks, Lucian
@Lucian: Yes it will affect things because you will be mixing frame rates. E.g., 29.97 from the EX1 and 30 from the 5D. I've never had any success with mixing frame rates in FCP (even though it is supposed to work) so I convert the frame rate to a common value when transcoding to ProRes. Compressor has a number of options for doing that, as very nicely described in Ken Stone's blog here.
Bruce, fabulous workflow. running smooth as can be.
One question, i used the downloaded custom setting for the 5D and when I put the converted clips into the sequence, FCP (7.0.1) said the format does not match. went ahead anyway and it works beautifully. is that normal?, or did I mess something up? Think mit could be that the 5D has 44 khz sound and the sequence was 48.
rgds and thanks Greg Marinovich
@Greg: Yes I think it is just the audio sample rate difference. Ignore the message and carry on.
Post a Comment