dual sound necessary if no back up required?

[eco_bach]

I understand the main problem with using the mic input in DSLR's is the preamps are s*%tty. But if the problem of AGC is circumvented and you use either an external pre amp or even the preamps on an external recorder, feeding the line out signal to your DSLR, then why is dual sound still necessary?

Does it have something to do with the bit rate, ie downsampling that occurs with the audio signal inside the DSLR?

I perhaps need to do my own tests, but skeptical that there would be any noticeable difference, especially for vocals recorded for interviews.

ie quality lav into a ZoomH6
line out from H6 into DSLR mic input (agc disabled)

compare audio recorded by Zoom with same audio received from Zoom in DSLR

 

[Alex H.]

AGC isn't the only issue. It just makes the other issues even worse.

The pre-amps are terrible, but there are some folks who decide it's perfectly fine to take the camera's input gain down to 1 notch above all the way while feeding a really hot signal to the camera. Sure, that minimizes the work (and added noise) of the camera's pre-amp, but it presents another big problem. As long as the person delivering lines is speaking in monotone and has no fluxuation in volume, it may be passable, but there's absolutely no headroom left on the input. Distortion/clipping can happen at this point, and it will not show up on the meters. This is what proper gain staging is all about. Plus, this kind of setup often results in recorded levels in-camera that have to be boosted in post, which pretty much negates any perceived benefits since that just amplifies noise.

The audio coming into the camera is compressed into the h.264 stream. Sound recorded on an external, dedicated sound recorder is (rather, should be) recorded at an uncompressed, 24-bit/48k WAV format. The compressed audio will not hold up to audio post-production like the WAV files will. That may not be an issue if there's little done to the dialog in post, but add some compression, EQ, etc. and you'll start losing quality.

Last, the DSLR likely offers no reasonable solution for headphone monitoring. Again, since clipping can easily occur on the input and will not affect the actual recording level, there would be no way to know when it happens without monitoring the recorded audio. It may sound clean in the pre-amp/recorder that's feeding the camera, but past that you won't know until it's too late. Further, many DSLR cameras do not display audio meters while recording. That's flying blind AND without a net.

That's not to mention all the other extra and more accurate tools audio recorders have that DSLR cameras do not.

 

[Richard Crowley]

There appear to be NO cameras at ANY price with good audio recording performance. From the very lowest end toy consumer products to the most esoteric top end cameras that cost more than your car.

At the low end, this is because the audio functionality is only there so they can sell their cheap cameras. The video part may not be much better.

In the middle, certainly there are professional broadcast-quality cameras with XLR inputs, decent preamps, metering, monitoring, etc. But the preamps are no better than your average low-to middle portable digital audio recorder. And with the added disadvantage that the person operating the camera is probably concentrating on getting good video as his primary task, and attention to recording audio is secondary at best.

Even if the audio performance of the camera is equivalent to your audio recorder, the advantage to using separate audio recording is that you can have a person (hopefully experienced in audio) actually watching the meters, listening to the audio on good headphones, and adjusting the levels, etc. This is effectively impossible when recording on the camera. Even a big camera with proper meters, knobs, and monitoring is busy getting good video angles, and is not in a good position to do anything decent with audio.

And at the very top end, many cameras didn't record audio at all. It was just assumed that if you were playing in that league, you were using double-system sound. And even high-end cameras that DO record audio have significant flaws in the execution that make it appear that they had no adult-supervision when it came to designing the audio parts. They were so intent on making the best camera in the world. they ran out of time or energy when they got near the end and somebody said "but what about audio?"

 

[Bruce Watson]

I understand the main problem with using the mic input in DSLR's is the preamps are s*%tty.

Not exactly. The problem with DSLR audio is that the entire audio chain is s*%tty. DSLRs are first and foremost still photography cameras. The vast majority of the design work is in getting the biggest-bang-for-the-buck still image quality. Adding video was, and is, an afterthought -- a feature used to sell still cameras. And if video is an afterthought, audio is almost a non-thought. The entire audio chain in a DSLR is probably way less than 0.01% of the cost of the DSLR.

You're much better off if you can feed your camera an actual line-out signal from a decent field mixer, over XLR cables. This would be my absolute minimum. Absolute. If you can't do that, send that line-out signal to a decent field recorder and do double system. It's not that hard to sync sound in your NLE, and the improvement is worth it -- if not to you, to your viewers / listeners. People will forgive all kinds of video flaws if you are telling a compelling story, but almost no one will forgive crappy audio.

 

[Andrius Simutis]

No offense to Richard and the other golden eared audio guys, but you also have to consider the balance between your workflow and the end audio quality needed. …in other words, what's good enough? I get the impression from too many audio guys that nothing is ever good enough, it's just an endless pissing match to achieve that last fraction of a percent difference that most people can't even hear. Video guys can do the same, but in the end we're not all shooting on Alexas or declaring that anything short of a RED epic 4k is complete crap.
Don't record or deliver crap audio obviously, but what kind of an end product are you needing? A wedding video where the music will be 90% of the end mix? Interview with no other sound? Feature film distribution? Corporate web video?

 

[Richard Crowley]

I deeply resent Mr. Simutis calling me a "golden ear"! :happy: I have fundamental differences with the REAL "golden ears" and will never agree with them.
I completely agree with Andrius that "good enough" is typically actually good enough.
But audio from ANY kind of DSLR is rarely "good enough" except for the lowest-priority stuff (ambient sounds, etc.)
And even if the audio chain in a DSLR were PERFECT, there would still be the problem that it would be almost impossible to monitor and control properly.
For that matter, it is difficult to monitor and control the VIDEO when shooing with a DSLR. (At least IMHO).
As Mr. Watson said, they are first of all a STILL CAMERA, with video thrown in as a low-cost "bonus". (Low cost to make, not necessarily reflected in the retail price).

 

[Noiz2]

Mr. Simutis's point seems to be if you set your bar low enough just about anything will work. And I actually agree with that. Also sometimes the bar can be pretty low. I think his point kind of misses on this list since I have never EVER heard anyone no mater what their ears were made of say you had to record to a Nagra, Zaxcom, Etc. When you start seeing folks recommend sound equipment that is more than what posters have already spent on camera gear give me a call and I'll buy you a beer.

I know my ears aren't golden because they tarnish.

 

[Andrius Simutis]

Mr. Simutis's point seems to be if you set your bar low enough just about anything will work…

Mr? Okay…didn't realize we were getting formal around here. Anyway, the point isn't to make garbage audio, but that we have to be realistic about what we're trying to achieve. There are a lot of situations where a Rode Video Mic plugged into the camera is just fine, and plenty more where it's not. The trick is to know the difference. I'm not going to hire a sound tech to come out and hike up a mountain while I film a snowboarder jumping off a cliff, but if I'm doing a short film with a lot of dialogue? Sure.

 

[Richard Crowley]

Even the professionals don't hire a sound guy when they are shooting people skiing/snowboarding down the mountain.
The difference is, they don't even bother with little toy plastic microphones mounted on their cameras.
The audio track you hear in the finished piece was created in post-production editing.

 

[Noiz2]

Guys, the options here are between the OP going into a cheap handheld or direct into a DSLR. So we are not talking about expensive equipment, or sound crews. The question was why bother with a recorder.
The answer is QUALITY.
A major step up in quality potentially (you can always screw up the sound).

No crew, NO expensive equipment.

And yes on a film with a budget all that snow boarding would be created in post because it doesn't mater what you do in production it's going to sound like garbage. But that is Andrius's problem not the OP's.

 

[Andrius Simutis]

The question was why bother with a recorder.
The answer is QUALITY.

And the counter answer is TIME. If you're turning it around quickly and the audio isn't the key element in the final product, then you'll save time without messing with dual system sound on the shoot and synching in post. That is if the audio isn't as important.

And yes on a film with a budget all that snow boarding would be created in post because it doesn't mater what you do in production it's going to sound like garbage. But that is Andrius's problem not the OP's.

Not always. Since nobody's shooting these movies on 16mm anymore the sound is recorded on camera in the field. Some of them will add sfx in post, but it's expensive and too often it's heavy handed and debatable if it really adds anything to the final product. Most producers will either leave ambient audio out entirely or just use what the on camera mic picked up on the hill. These aren't hollywood movies and the audiences are perfectly happy with the audio that was captured by those "inferior" means.

 

[Richard Crowley]

It is true, "Reality" shows and MTV "Shaky-cam" have dramatically lowered the public's expectations of quality. But even "Blair Witch Project" had all the original sound track discarded and completely (and professionally) re-created in Foley and ADR. People who pay for tickets to sit and watch the Big Screen apparently will put up with lousy video, but they draw the line at lousy audio.

Audio without video was "radio" for decades before TV came along. (And still a very popular medium.)
Video without audio is called "dead air" and alarms go off at the TV station.
If you don't get the audio restored in a fraction of a minute, your audience has all moved to the next channel.

 

[TheDingo]

there are some folks who decide it's perfectly fine to take the camera's input gain down to 1 notch above all the way while feeding a really hot signal to the camera. Sure, that minimizes the work (and added noise) of the camera's pre-amp, but it presents another big problem. As long as the person delivering lines is speaking in monotone and has no fluxuation in volume, it may be passable, but there's absolutely no headroom left on the input.

This is not the case with the Panasonic GH3 camera, where you have the usual 12 dB of audio headroom plus the built-in limiter circuit if the signal gets really hot. ( the older GH2 audio circuitry worked this way too )

Distortion/clipping can happen at this point, and it will not show up on the meters.

...But you will hear it if you monitor your audio via the camera's headphone jack. Over the past 3 years I've never had clipped or distorted audio when recording with the GH2 / GH3 cameras.

The audio coming into the camera is compressed into the h.264 stream

The Panasonic GH3 camera can be set to record audio in uncompressed LPCM format, which is a nice improvement over the GH2. ( along with the headphone jack for monitoring )

Last, the DSLR likely offers no reasonable solution for headphone monitoring

Some cameras do. ( i.e. Panasonic GH3, Nikon D7100, Canon 5D Mk3, etc... )


...With the proper audio support gear, DSLR audio can be pretty good, but external double-system setups will always be better. I use GH3 camera audio for smaller, lower budget jobs, and a double-system setup for work where audio is a much bigger deal. I expect the new GH4 will have better audio than the GH3, which is already pretty good for in-camera audio.

 

[Alex H.]

This is not the case with the Panasonic GH3 camera, where you have the usual 12 dB of audio headroom plus the built-in limiter circuit if the signal gets really hot. ( the older GH2 audio circuitry worked this way too )

...But you will hear it if you monitor your audio via the camera's headphone jack. Over the past 3 years I've never had clipped or distorted audio when recording with the GH2 / GH3 cameras.

I disagree. That -12dB headroom is what you're seeing on the meters. What I'm talking about is not related to the metered recording levels, but to the level of the electrical (analog) signal at the input of the camera, before the pre-amp. If the only way to get a manageable, and relatively clean, signal recorded in-camera is to turn the input gain all the way down, it means that the level of the incoming signal is too hot... basically feeding a line-level signal to a mic-level input. There is a very real risk of distortion/clipping in the pre-amp in this case, which will not spike the meters.

This is about gain staging, or rather proper gain staging. In a signal chain that has several stops along the way, there are many opportunities for damage to the audio signal that will have no impact on the final, metered recording level.

If you happen to have one of the still-uncommon models of DSLR that offers headphone monitoring, more power to you. I never asserted those didn't exist, but there is a very large user base out there of DSLRs that do not have monitoring, or don't have it without a workaround.

The Panasonic GH3 camera can be set to record audio in uncompressed LPCM format, which is a nice improvement over the GH2.

Excellent. Still the exception rather than the rule, but the uncompressed format isn't going to improve the shoddy audio circuitry that comes before it. In fact, the better the recording quality becomes at the end of the line, the more it will reveal imperfections upstream.

...With the proper audio support gear, DSLR audio can be pretty good, but external double-system setups will always be better. I use GH3 camera audio for smaller, lower budget jobs, and a double-system setup for work where audio is a much bigger deal. I expect the new GH4 will have better audio than the GH3, which is already pretty good for in-camera audio.

Sure... manufacturers are starting to take in-camera DSLR audio a little more (relatively) seriously, but while it may be passable for a no-budget/low-expectation production it will never match what can be captured with a dedicated audio recording system. Even the highest-end cameras only have sound a a secondary thought, and a Sound Devices recorder is going to outperform an Alexa or RED. We're talking lower-budget here, I get that, but the point is that in-camera sound is never an R&D priority on any level with any manufacturer.

But we have gotten way off from the OP, and aren't even certain which DSLR is being used. With the Zoom involved in the signal chain in both scenarios eco_bach presented (it's just a question of whether to record only in-camera or with the Zoom as well), it makes no sense not to record with the Zoom. Even if it is still feeding the camera, and the camera's internal audio is deemed to be "good enough" for this project, having the option to go back to the external recorder is a much better safety net.

 

[TheDingo]

If the only way to get a manageable, and relatively clean, signal recorded in-camera is to turn the input gain all the way down, it means that the level of the incoming signal is too hot...

Normally this would be true, but with the Panasonic GH2/GH3 cameras we are turning down the input gain is to reduce the effect of the built-in AGC circuitry. So it's not an issue of having a very "hot" signal, but rather Panasonic not giving us the ability to shut off the AGC circuit which ruins the noise-floor in quiet recordings, so by lowering the camera's input gain we can come close to eliminating the problematic AGC issue. ( at the lowest camera gain setting the AGC has a +3 dB effect, where at higher gain settings the AGC can boost the audio levels by +25 dB )

Sure... manufacturers are starting to take in-camera DSLR audio a little more (relatively) seriously, but while it may be passable for a no-budget/low-expectation production it will never match what can be captured with a dedicated audio recording system.

Totally agree with you here. Nothing beats dual-system sound, but for many video projects the in-camera audio is more than good enough if you use good audio support gear with your camera.

 

[Alex H.]

Normally this would be true, but with the Panasonic GH2/GH3 cameras we are turning down the input gain is to reduce the effect of the built-in AGC circuitry. So it's not an issue of having a very "hot" signal, but rather Panasonic not giving us the ability to shut off the AGC circuit which ruins the noise-floor in quiet recordings, so by lowering the camera's input gain we can come close to eliminating the problematic AGC issue. ( at the lowest camera gain setting the AGC has a +3 dB effect, where at higher gain settings the AGC can boost the audio levels by +25 dB )

Non-defeatable AGC negates any benefits of uncompressed audio.

Okay, so you're lowering the input gain to minimize the effects of the AGC. That gain still has to be made up somewhere. It can be made up by pumping a hotter signal into the camera, which is common, and that risks overmodulating the input as I have previously stated. The gain can also be made up after the fact, whether there is a secondary level control in-camera or it is simply brought up in post, in which case noise floor is increased with the signal. The former can work, tenuously, but leaves little wiggle room. The latter increases noise.

 

[eco_bach]

Alex, the firmware update for my 1DC is supposed to offer a true line level input now, so I'm hoping that alleviates some of the issues.
Still dealing with COMPRESSED 16bit audio, so little room for any work in post.
For simple dialogue in interviews, I think the quality will be 'ok'. Need to do some tests.

 

[Richard Crowley]

Alex, the firmware update for my 1DC is supposed to offer a true line level input now, so I'm hoping that alleviates some of the issues.

That seems impossible. How do you bypass the (noisy) mic preamp in firmware? Someone is either misinformed or misspoke.

Still dealing with COMPRESSED 16bit audio, so little room for any work in post.
For simple dialogue in interviews, I think the quality will be 'ok'. Need to do some tests.

It is not just the "quality". It is also an issue of compound artifacts in distribution compression as a result of compression -> decompression -> editing -> re-compression.

 

[Noiz2]

And the counter answer is TIME. If you're turning it around quickly and the audio isn't the key element in the final product, then you'll save time without messing with dual system sound on the shoot and synching in post. That is if the audio isn't as important.

I do believe that sums up my lowering the bar nicely. IF you don't care about the sound you can do anything. IF you do then any time saved by not syncing is lost when post tries to make it usable.

Not always. Since nobody's shooting these movies on 16mm anymore the sound is recorded on camera in the field. Some of them will add sfx in post, but it's expensive and too often it's heavy handed and debatable if it really adds anything to the final product. Most producers will either leave ambient audio out entirely or just use what the on camera mic picked up on the hill. These aren't hollywood movies and the audiences are perfectly happy with the audio that was captured by those "inferior" means.

I probably was not clear enough with "any film with a budget". Any film that has a full paid crew and is going through the workflow for distribution with an actual distributor will create it all in post. I don't include YouTube films, and reality TV and such in that group. Some are well made but they are not "films with a budget". Films with budgets haven't shot 16mm in a LONG time. Supper 16 from time to time, but actual 16 doesn't blow up well to 35mm and 16mm distribution has god-awful optical track sound.

Also shooting double system and doing post sound are not at all connected.

 

[eco_bach]

That seems impossible. How do you bypass the (noisy) mic preamp in firmware? Someone is either misinformed or misspoke.

http://www.usa.canon.com/cusa/professional/standard_display/cinema-firmware-1dc

The EOS-1D C now accepts a line-level audio input via its mini 3.5mm connector, for users who want to use an optional outboard mixing console, for example, or a high-output microphone. In Line Input mode the standard input level is -8 dBV* with peaks up to +6 dBV. To accommodate different external sources, sound-recording levels are adjustable at 64 step levels