April 30, 2012

The Battle of the Preamps: 5D Mark III vs. D800 Audio Recording

Audio recording internally to DSLRs has been mediocre at best. It's great for scratch audio when you're doing dual-system sound, but for the most part, it's a real pain. I've been testing the 5D Mark III and the D800, but one of the tests I wasn't able to do as thoroughly as I wanted was to test the internal audio recording of both cameras with a proper microphone. I know that many out there would ask why you'd ever plug directly into the DSLR without some other external preamp box, but sometimes (like at NAB), having the least amount of equipment that can fail is best. I met Dave Dugdale at NAB, and in this video he takes the time to test out both the Nikon D800 and the Canon 5D Mark III for the quality of their internal audio recordings.

This video was originally posted over at Dave's site, Learning DSLR Video:

So who wins? Well, neither one really, if you're comparing to an external recorder - even a cheap one like the Zoom H1. I think it's possible that the 5D Mark III might have a slight advantage in noise, but it's really a negligible difference - I'd be more likely to call it a draw.

While the actual process of recording audio has improved for this generation of cameras, with the Mark III actually able to adjust levels while recording, audio quality hasn't made any great leaps over the last generation. The D800 is unable to adjust levels while recording, which I found during my testing, but that could be forgiven if the quality of the audio internally was spectacular. Unfortunately, it seems both cameras still leave a lot to be desired, and the noise of the preamps in each DSLR is well above the noise of an external recorder like the Zoom H1 (and there are some other good recorders that will give even better results).

My suggestion, if you don't want to use an external box of any kind, is to record the levels as high as possible without clipping. This can be dangerous when the levels cannot be adjusted while recording, but it will be the best way to get the cleanest audio possible internally. If you don't do this and you have to bring up the levels in post - it's going to get messy. The idea behind doing this is that by bringing up the noise floor, once you're in post you shouldn't have to raise any levels - in fact - you'll be bringing them down and you will get nice, clear audio. This will minimize the amount of noise introduced into the post process. If you aren't adjusting levels in post, however, this process won't necessarily give you better sounding audio.

Either way, I think we've seen just about the highest level of audio recording we will probably get out of these cameras, since it's much more likely that Canon will concentrate their resources on large sensor video cameras, and Nikon may very well do the same. I've gotten a tremendous amount of feedback on my Mark II/D800 candlelight test, and my goal will be to do an updated version that compares them at equal brightness, with color correction and noise reduction applied in post - so stay tuned for that test.

[via Learning DSLR Video]

Your Comment

14 Comments

All valid points but at least the two cameras tested record PCM WAVE audio not compressed audio, as the AVCHD-based cameras (GH2, Sony A77) do.
Also mic choice can make a difference: a mic that sounds good with great pre-amps might not sound as good on a camera with less effective audio inputs. I've had good results from the D7000 using Nikon's little ME-1 mic, in the right circumstances, but an expensive Sennheiser would be drowned by the pres. A mic with its own power, eg the Rode NTG2, can also help as the camera's pres aren't working as hard.
I'd not record, say, music that way but both are good enough for vox pops.

April 30, 2012

0
Reply

Hey Joe, nice to meet you at NAB, and thanks for posting my video here!

April 30, 2012

-1
Reply

Great meeting you too! Absolutely, I thought it was a great way to test both cameras.

April 30, 2012

0
Reply
avatar
Joe Marine
Editor-at-Large
Shooter/Writer/Director

I use a Juicedlink box to pump audio into my 5D mk2. I've got the input volume on the camera only one notch above zero so that the Juicedlink pre-amps do all the work. The audio I get sounds pretty good to me and no syncing in post.

May 1, 2012

0
Reply
Stu Mannion

I'd be very careful about suggesting recording at the highest level without clipping. That's fine with limited or predictable dynamic range, but when the sound goes beyond that zone, it'll clip, distort, or brickwall, resulting in unusable audio.

I think the caveat is most people want good wave forms from the onset to edit in an NLE. I almost never rely on a video NLE to increase levels. I prefer to record cautiously (aka low) and allow headroom, usually a lot of headroom. Preamps I've heard in camcorders and external recorders such as the H1 don't have much noise floor. Current technology is decent, especially if signal-to-noise ratio is high, even if the waveforms are low, and spikes will not distort (I'm talkin' 10-20 db headroom).

My workflow is to foremost process the audio in Amadeus 2 (on a Mac), and increase the levels there. The app opens mov files and will save as mov or aiff. Then bring that audio into an NLE. The alternative method is to place the mov on the NLE timeline, and from there export an AAF to Pro Tools and use track, bus, and a plug-in to increase the level. That's the long way. Hence, I almost always use Amadeus.

Trying to create thick waveforms on-the-fly while shooting is precarious, especially with run-n-gun. When I do sound for DPs and they complain about the video audio being too low, I'll send them the Zoom audio that I processed in Amadeus and they are happy. I never tell them I increased the level in Amadeus (it's too much information for a video editor) and they never hear a noise floor, and they don't get clipped sound either.

May 1, 2012

1
Reply
animal_264

i think the rode video mic is a piece of shit. i have it myself and get the exact same hiss problems. my mate has got a sennheiser dslr mic and it sounds perfect. almost no hiss at all. i think the rode is just a bad call in any situation.

sorry. thats my findings from at least 10+ interview shoots.

May 1, 2012

0
Reply
Pat

On 5D hit info button and you've got levels on screen fulltime

May 3, 2012

0
Reply
JR

I don't get any benefit from this test or may be I do not have such sound knowledge to see the difference. So, what I want to ask is: could you give us clear and more tangible tests that could benefit to deciding which one to buy or how to fix the defects.

May 3, 2012

0
Reply
nyuntshwe

One more thought: I don't think only the sound quality alone would not decide which camera is more suited for you. So, the tests should be more inclusive.

May 3, 2012

1
Reply
nyuntshwe

You should check out the ongoing review we are doing here at NoFilmSchool:

http://nofilmschool.com/tag/markiiid800review/

May 3, 2012

1
Reply
avatar
Joe Marine
Editor-at-Large
Shooter/Writer/Director

I use wireless lav mics always when connecting directly to the DSLR.

May 4, 2012

1
Reply
Cal

Any preamp will really clean up the audio. I was hopeful with the D800 it would be possible to skip and go xlr->1/8" but a Beachtek preamp or even a mono headphone preamp really really cleans up the hiss and lowers the noise floor to a very clean level. Also using an external preamp on the D800 allows you to change level even while recording.

May 4, 2012

0
Reply
Nate

does the magic lantern hack on the 5DII help at all with the audio quality? I know it has lots of added audio controls. But does anyone out there use it instead of recording separately and get good quality results?

May 7, 2012

0
Reply