[Joel Ninmann]

What model Zoom recorder did you use? The H4N? What frame rate did you shoot the video on the 5D Mk II at? 24 fps (actually 23.976 with firmware version 2.0.4) or 30 fps (actually 29.97 with firmware version 2.0.4)? What firmware version is installed in your 5D? What editing software and version are you using (Final Cut Pro X, Final Cut Pro 7, Adobe After Effects, Avid)?

Generally the quick solution (without getting into why) is to set your audio down to a speed of 99.9%. Try that.

Joel Ninmann
Instructional Specialist
Department of Communication Arts
University of Wisconsin-Madison

[Joel Ninmann]

Justin,

I think the link you posted accurately sums up what you should do when a slate is blurred. Personally, I am with David — I sync to the last blurred frame of the slate. This doesn’t mean we are right, though.

Thanks for letting us know you’re seeing the same issue with a Canon T2i. This means we now have 3 HDSLR cameras that all see this issue — the 5D Mark II, 7D, and T2i. Anyone seeing this issue with a Nikon HDSLR?

Another question, given that Final Cut Pro X was released today:

Does the automatic audio syncing on import option in FCPX sync HDSLR on-board audio correctly, or do we still see the same 2-3 frame sync issue? Could someone with FCPX installed test this out by importing some footage from any of the 3 above cameras and let us know? As soon as outside forces permit, I will do so myself, but that may be a few days or weeks. I’m sure lots of people on this board would appreciate someone else being able to do so sooner.

I’m glad more and more people are stumbling upon and waking up to this issue. 2-3 frames is completely unacceptable, and I STILL shudder to think about the thousands of HDSLR shooters out there syncing via software like Pluraleyes that are missing this issue with ALL their footage on a daily basis.

[Joel Ninmann]

Good to see some people are still aware of/trying to come up with a solution to this problem. Also good to hear from someone using a 7D with the same problem. Now we know it isn’t just the 5D Mark II. I suspect 48.000 has something to do with the issue, but since these cameras do not give you the option of changing sampling rate, I have no way of testing this.

My video clap frames have been consistently 2 frames off. I have not seen many that are only 1. I also am fairly certain that shutter speed does not have an effect on this issue, especially when you consider that when shooting video on these DSLRs, the shutter is not a true mechanical shutter, but a fake/digital shutter. In my experience, changing the digital shutter speed has had no affect on this audio sync issue, but I invite people to test it out for themselves to double-check my experience.

Even if it WERE due to shutter speed, this wouldn’t matter, because you almost never want to have a DSLR’s shutter speed (when shooting video) set to anything other than 1/30 or 1/60 (for NTSC) or 1/50 (for PAL) due to monitor/light strobing that you can get in the opposite region otherwise. Also, its been my experience that longer shutter speeds’ (e.g. 1/30) blurring effect is not a true motion blur (due to the aforementioned lack of a physical/mechanical shutter), but rather a largely displeasing-looking digital blur.

No thank you to any shutter speed other than 1/60 for NTSC.

[Joel Ninmann]

David,

I don’t think its from Final Cut Pro, though FCP does process video and audio slightly differently. Obviously, FCP only deals with video at a maximum of the frame-by-frame level. Audio, however, can be zoomed-in to the sub-frame level.

Still, I can’t be sure that this issue isn’t caused by FCP unless it is tested in a different editing program. Maybe someone out there with Avid or Premiere could do so?

I can try and wrangle a copy of Premiere around here, but it may take a little while. I’ll let you know if I do.

[Joel Ninmann]

*** UPDATE ***

I have tested this video delay with a Fostex PD-4 DAT tape recorder. The Fostex will play a reference tone when you flip a switch to the right. I set this reference tone to a level of -12 dB and the tone played through the Fostex’s on-board speakers. Once again, if you play the file back using an editing program that allows you to playback at the frame-by-frame level, you will notice the video delayed by approximately 2 to 3 frames. The sound of me flipping the switch on the Fostex is heard along with the start of the reference tone on frame 1, then on frame 3 or 4 the LEDs of the level bars begin to materialize on the Fostex’s monitor screen.

Can we honestly simply chalk this up to the 5D processing the video and audio at different rates? I know the 5D was not designed and built to be a professional video camera with timecode and everything, but if Canon is marketing this thing to professionals (as I’ve seen in their commercials), then they need to address the lack of on-board reference audio sync. What does everyone else think? Can someone test this with a 7D at 720/60P?

I recorded the below file at a distance of about 2 feet with the Canon 5D Mark II (latest firmware 2.0.8) with a 50mm lens. Shutter was 1/60. Audio was set to manual and levels set to -12 dB. I transcoded the files from the camera to Apple ProRes 422 via Final Cut Pro 7.0.3 and the Canon EOS movie plugin E1. The file below was compressed via Apple Compressor to H.264 1500 kbps 1920×1080 29.97P for video, and 48.000 256 kbps audio.

https://dl.dropbox.com/u/11690716/fostex-sync-test.mov

[Joel Ninmann]

Make sure your shooting mode is set to “Single” instead of “Continuous”. Look up how to do this in the manual if it isn’t immediately apparent.