in cam audio with 5Dm2, FP33

[Josh Bass]

I'm preparing to shoot a short film, and I know the conventional wisdom is to use an external recorder for audio when shooting with one of the video DSLRs, but this is a low/no budget production, and neither myself nor the director has access to such a device, we won't be purchasing one any time soon.

The options are in camera audio on the 5D, or using my XL2 as an audio recorder (my original plan).

reading on here, I see most people say recording dialogue is okay with the 5D, despite it's high pass filter.

I've been doing some testing, and am not pleased with the results. Before I give up and go to the XL2, I want to make sure I'm doing this right.

So 5D mic input is mic level, right?

So I'm running an AKG blueline hypercardioid into an FP33, coming out mic level via an XLR to 1/8" adapter cable to the 5D. I run tone from the FP33, setting master on the FP33 so the tone reads 0 on its meter. I then set the camera's meter exactly halfway (I'm guessing here, but if the highest notch on the meter is 0, and the lowest -40, exactly between them should be -20, right?). This seems like it should be the proper gain structure ( to me, anyway. . .unless something I did above is wrong), but when I did a test recording on the camera, with peaks about -10, the resulting recording sounds pretty hissy. . .FCPs meter tells me the hiss level is about -36. That is way loud.

So, did I set something up here wrong, or is this just how it is on the 5D?

[Daniel Epstein]

Josh,
I am not sure what your particular problem is in getting the recording on the 5D to sound decent but you should try recording the signal a little hotter than midway on the inputs and see if that helps your gain structure on the camera and then run the mixer levels a little lower to see if that helps as well.
I did a shoot a few months ago using a 5D with a mixer on mic level and back up recording to a zoom and the editor said there was very little difference in the quality (which he was happy with) of the camera tracks and the recorder tracks with the recorder winning by being a bit warmer on the low end.

[Josh Bass]

I ran TONE at about halfway on the camera's meters (what I assumed to be -20, the way you would do it on a proper camcorder), but when actually recording peaks were about -12 to -10.

[Steve House]

Originally Posted by Josh Bass

I ran TONE at about halfway on the camera's meters (what I assumed to be -20, the way you would do it on a proper camcorder), but when actually recording peaks were about -12 to -10.

That's about what you'd expect. Due to the nature of the mathematics of sine waves, a given level tone will read the same whether the meter is peak reading or averaging. Real-world sound such as speech, OTOH, reading a given level on an averaging meter would read 8 to 10 dB higher on a peak-reading meter. So sending tone reading 0VU on the mixer to the camera, adjusting the camera input level so the tone reads -20 dBFS on the camera meter means that speech reading 0VU on the mixer's averaging meters will read about -12 dBFS on the camera's. You could set the camera input higher but remember you have to avoid peaks ever hitting 0 on the camera meter at any cost.

[Neil Hurley]

Hi Josh,
you may find some useful information at this link.Good luck.
Neil

Audio Performance of Canon 5D Mark II Camera|Sound Notes|Sound Devices, LLC

[Neil Hurley]

Ha, sorry!! this one :Audio Performance of Canon 5D Mark II Camera|Sound Notes|Sound Devices, LLC

[Josh Bass]

Okay, looks like that link advises simply running camera's audio at its absolute lowest setting, one click away from totally off.

They then say set mixer's limiters to just below 0 dBFS.

Don't really have this option on an FP33 (I think the factory limiter settings are close to 1 or 3 db below its max --- +20 by default)

So then you don't line up tone at all? Just set mixer master to 0 and adjust individual mixer channel pots 'til it reads at -12/10 for peaks on cam meters?

[Neil Hurley]

"So then you don't line up tone at all? Just set mixer master to 0 and adjust individual mixer channel pots 'til it reads at -12/10 for peaks on cam meters?"

Yes,that seems to be what they are saying. I don't own an SD mixer and I have not tryed to record direct to the 5D so I can"t say for sure if it will work.Id say its worth doing some tests using the above method and your own mixer.As I understand it the idea is to use more of the the mixer gain and less of the camera gain.

Good luck,Neil
__________________

[Josh Bass]

I will try it. Still wondering if tape outs (-10) are too powerful for 5D, if it HAS to be mic level.

[Steve Oakley]

before you say you can't afford even a used zoom h4n.... then you probably can't afford to feed the crew or talent, pay for gas, props, locations, a drive to hold the media, another drive to back it up. given all that it takes to do something like this, doing audio poorly will mean just one thing - either it will kill the project because no one will be able to deal with ADR, it will die because its audio is unusuable, or there will be so much more extra work after the fact people will just quit.

put another way, for every $1 you save doing things the wrong way in production, expect it will cost $5-$10X each dollar saved to fix it in post, if it can be fixed. finding a couple hundred for a modest recorder needs to be on your list, and really there isn't any way around it.

[Josh Bass]

I appreciate that, but be that as it may, it's still not gonna happen.

[Steve House]

Originally Posted by Josh Bass

...I run tone from the FP33, setting master on the FP33 so the tone reads 0 on its meter. I then set the camera's meter exactly halfway (I'm guessing here, but if the highest notch on the meter is 0, and the lowest -40, exactly between them should be -20, right?). This seems like it should be the proper gain structure ( to me, anyway. . .unless something I did above is wrong), but when I did a test recording on the camera, with peaks about -10, the resulting recording sounds pretty hissy. . .FCPs meter tells me the hiss level is about -36. That is way loud.

So, did I set something up here wrong, or is this just how it is on the 5D?

Remember that the decibel scale is logarithmic, not linear. The -20dBFS point is not halfway between the -40 dBFS and 0 points, it's closer to about 1/3 the way down from zero.

[Josh Bass]

Ah! I know in practice it's not linear, but some meters don't reflect that. Good to know. If that SD article is right, though, it sadly doesn't matter.

[Steve House]

Originally Posted by Josh Bass

I appreciate that, but be that as it may, it's still not gonna happen.

I don't get it. If there budget isn't there to shoot it right, why not defer production a little bit and build the budget to the level necessary to do the job right rather than proceeding to waste the time and what budget you do have on a film that'll sound so bad no one will want to watch it? Doesn't seem like a good use of time and resources to me.

[Chris Barcellos]

Frankly, not be able to monitor the actual recording as it is going to camera, is a problem for you. With Magic Lantern, there is a resolution to that, but at this point, it does not work with the latest version of the camera's firmware. If you have firmware 2.04 still in camera, it will.

Thus, it might be just as easy to record to the XL2, via a method you are familiar with, then line up the sound in post. To do that, set the camera sound on Auto Gain and record through the camera mic for a tract that can be used to match the sound.

Plural Eyes has a 30 day trial that you can use to line match you XL2 sound with your footage in common NLE's.

Depending on your NLE, look at Dual Eyes, from the same manufacturer, as it is stand alone, and can batch process and actually end up with a file that has the separately recorded should attached to it.