Best Solution For Canon 7D? Thinking About Taking It Back...

Pedanes Bol · March 6th, 2010, 07:37 PM

Quote:

Originally Posted by Alex Donkle

On top of that, although people mention 48k, there's a lack of into about what bit-depth the camera uses. It's either 12-bit or 16-bit. 16-bit is okay, but with AGC a 12-bit recorder could easily be what Canon used. No one seems to know though and Canon hasn't posted specs that I've seen.

Quicktime inspector shows 16-bit:

H.264, 1920 x 1080, Millions
16-bit Integer (Little Endian), Stereo (L R), 48.000 kHz

Jon Fairhurst · March 6th, 2010, 07:50 PM

Yeah,

I'm that guy. :)

Overall, the juilcedLink into the 5D2 is surprisingly good. The H4n is nearly as good. The Microtrack II is not so good - unless you drive it from external preamps, and even then it has audible flaws.

Overall, the JL/5D2/ML combo is VERY clean, and amazingly crisp, but slightly harsh. The H4n is about 6dB noisier and slightly soft or mushy by comparison. Both give professional, but not audiophile, quality.

juicedLink now makes the DN101 for fooling the auto-gain. It also has a headphone output and meters. I haven't had a chance to test it yet.

So, why is the JL/5D2/ML slightly harsh? I believe that I'm hearing some high frequency aliasing. When Canon updates the 5D2 from 44.1 to 48 kHz, this could help. Of course, the other Canon cams already do 48 kHz.

Why is the H4n a bit mushy? I believe that I'm hearing phase distortion. If the highs and lows are slightly delayed from one another, the result is a smeared edge. This isn't something that is fixed by simply cranking the HFs. Similarly, the JL/5D2/ML isn't crisp because the highs are boosted. The frequency range sounds pretty flat to me.

But the differences are slight enough not to worry. Both solutions will sound great for many applications. I've heard our JL/5D2/ML recordings in a theater (48 Hour Film Project), and the noise level was nice and low without any noise reduction being applied. Still, it was slightly harsh in that environment, but our sound quality was among the best of those who participated. I'd want better for a Hollywood release though...

Corey Benoit · March 6th, 2010, 08:25 PM

so a dn101 and a juicedlink cx231 combo with a studio recording condenser mic on a boom pole will sound good right?

will it blend with my ex1r? i will be recording from boompole into that camcorder and using the 7d for alternate scenes...

Jon Fairhurst · March 6th, 2010, 09:26 PM

Yeah, it should sound good, due to the clean preamp. I haven't tested it though, and I don't own a 7D, so I can't offer any details.

Marco Leavitt · March 7th, 2010, 12:32 AM

For those who are happy with the results you're getting, more power to you. For me, I would never trust a setup that relied on fooling the autogain of a camera with funky preamps for anything critical. Can you really be sure that you are always calibrated correctly? To say that you heard sound recorded to the camera in a theater and it sounded fine, well, I'm not sure that really tells us anything. If you record a nice strong signal to practically any device it can sound all right. It's what you don't know you are missing -- that's where the real loss is. But I would say, definitely, that if you concentrate on good recording techniques, with good mic placement, proper levels and gain staging, you can get surprisingly good results with the humblest of tools. So I'm not saying that audio recorded with the Juicedlink/7D combo is necessarily going to be bad. Just for me personally, it all just seems a little too wonky to trust. It would be really great if somebody were to actually conduct a test of the signal to noise ratio, frequency response and distortion of this combo. It would sure take a lot of the subjectivity out of the debate. There certainly do seem to be a lot of people who would like to record direct to that camera. Also, even with the cleaner preamps of the Juicedlink, you can't bypass the preamps in the 7D altogether. All you can do is force them down, and at a certain point you are increasing the distortion and possibly encountering an impedance mismatch. The sweet spot on the gain for most camcorders is usually about one third of the way up from the bottom. I don't know how you can be certain exactly what the internal preamps are doing on the 7D.

Corey Benoit · March 7th, 2010, 01:20 AM

recording directly to camera saves time, and is way more practical and convenient with todays technology...

the only reason that the film inustry recorded seperately was to get better sound qualtiy and didnt have good recording directly with the cameras...

todays cameras have the ability with technology to record high quality sound...

Jon Fairhurst · March 7th, 2010, 01:36 AM

Again, I don't have a 7D and I haven't tested the DN101.

[NOTE: I've edited the following for accuracy.]

I do have the 5D2 and have tested it with the juicedLink with Magic Lantern. I set the gain to +17 dB in analog and zero in digital, preserving the maximum dynamic range. With my specific test setup, noise hovered around -96 dB, which would be the theoretical limit of the 16-bit recording. The results were about 6dB better than the H4n.

And I don't know why the juicedLink would be a "funky" preamp. Okay, the name is kind of funky :), but this thing has some of the cleanest gain that you'll find. (I see you were talking about the Canon preamps, which aren't super clean on their own, but exhibit no bad behavior. Note that the Canon AGC is done digitally, not in the preamp.)

From my tests with the Beachtek and 5D2 with stock firmware, the signal jamming approach drives the digital gain down to or near zero. The problem was the analog section. It's set on the 5D2 to a fixed +31 dB. It's possible that even with a clean signal, the noise could be around 14 dB higher than on the 5D2/ML setup. It could be higher, due to bleed from the AGC defeat signal, but it really needs to be tested to know for sure.

It will be interesting to see how Canon programmed the analog gain in the new 5D2 firmware. They have one adjustment. Does it control the analog side, the digital side, or does it mix the two?

My feeling is that with the expected noise, some will be happy with that performance with the advantage of not having to sync. Others will prefer a double system. A third option is to use a field mixer into a top recorder and to feed the line level into a juicedLink and 7D as a clean sync track. This is especially good if you are using a wireless system and the 7D internal mic wouldn't be able to pick up the dialog. It also gives you a backup track, if something goes wrong with the primary recording.

Brian Luce · March 7th, 2010, 05:03 AM

Jon have you had a chance to play/compare with the Tascam DR100?
I exchanged my Zoom for the Tascam because I liked to features better, never tested the audio fidelity though.

Corey Benoit · March 7th, 2010, 05:32 AM

will this work with the 7d? it looks like it will....

http://www.bhphotovideo.com/c/used/531903/Cavision_RS1580S_SPE_RS1580S_SPE_Shoulder_Pad_with.html

Steve House · March 7th, 2010, 06:03 AM

Quote:

Originally Posted by Corey Benoit

recording directly to camera saves time, and is way more practical and convenient with todays technology...

the only reason that the film inustry recorded seperately was to get better sound qualtiy and didnt have good recording directly with the cameras...

todays cameras have the ability with technology to record high quality sound...

"Saving time" and "more practical and convenient" are valid criteria only if you do not have to compromise in any way on quality in order to obtain them. For the most part, today's cameras are able to record usable to good sound, that's true. But the objective isn't just getting usable sound, it's getting the very best sound possible under the conditions where you're shooting. A solution that compromises quality in order to achieve convenience is not an option, IMHO.

Corey Benoit · March 7th, 2010, 11:17 AM

also what is the best way to edit 7d footage in premiere pro cs4? i have avch and h.264 codecs and presets....the 7d records in .mov files, and its really slow...

heres my computer specs:

core i7 920 3.4ghz 8 mb cache

triple channel memory 3 x 1 gb ddr 3 @ 1.7 ghz

SLI 2x geforce 8800GT's both overclocked

WD hard drive 32mb cache

do i need more ram, my cards are pretty fast...

Marco Leavitt · March 7th, 2010, 11:32 AM

Jon, when I referred to funky preamps I was talking about the 7D, not the Juicedlink, which by all accounts has very nice preamps for the cost.

Regarding your tests, and I mean this respectfully, but I'm skeptical about your test results. If you are really getting 96 dB SNR with that system, then you are achieving better results than any camcorder, or even any 16 bit recording device that I'm aware of. In fact, I don't think that kind of performance is even possible on a 16 bit recorder. I think you need to review your workflow and software, and maybe test it on another device with published specs to see if you can replicate their results. I do very much respect that you are trying to go about this in a scientific way, so please don't feel that I am being combative. Even 75 dB SNR is quite good, and better than many professional camcorders with actual XLR line inputs.

Jon Fairhurst · March 7th, 2010, 01:35 PM

@Brian: I haven't had a chance to try the DR-100, but would love to. It gets consistently good reviews.

@Marco: You're right. I didn't do these tests with calibrated lab equipment. I did the test relatively. So I shouldn't claim absolute numbers for the units. My mistake.

But I did do the tests consistently from unit to unit. The nice thing about the dB scale is that it's relative. I first measured the sensitivity of each unit with a 500 Hz tone. I wasn't able to simply calibrate everything, specifically, because the Beachtek has no gain and it defeats the Camera AGC, so there's no gain adjustment there either. Similarly, the Microtrack II has no analog gain adjustment. The H4n and the JL/5D2/ML were the only units where I could boost gain in the analog domain.

My assumption is that if I used each of these systems with the same mic and performance, that I'd have to boost or lower the gain in post to achieve the desired mix. So I applied gain to my noise recordings based on the calibration levels that I had measured earlier.

The details of the noise test are here:

And you can download an uncompressed recording here: http://p3pictures.com/audio_5d2/Audi...art5_Noise.zip

Regardless of the numbers, the juicedLink clearly has the lowest noise - especially as we move up the frequency range. Frankly, the thing is eerily quiet, considering what I had expected plugging into a 1/8" jack on a still camera.

Watching the video again, I had forgotten that I ran the analog gain of the camera at +17 dB. (I've been running +10 dB these days.) So, with the 7D analog gain likely fixed at the maximum of +31 dB, I'd predict that the DN101 will have at least 14 dB more noise than my JL/5D2/ML test produced. It could be more, due to the bleed of the AGC jamming signal.

But again, I haven't tested it.

And thanks for reminding me that my tests were not absolute, but relative. I shouldn't have quoted the numbers as such.

Mestizo Devon · March 9th, 2010, 10:57 AM

Quote:

Originally Posted by Stephen Mick

Firmware upgrade is for 5DII, not the 7D.

Contact canon for these features to be added as well, but you gotta love software limitations!!

Laurence Kingston · March 9th, 2010, 11:28 PM

Here's a completely different way to do it if you are working with Sony Vegas or FCP. There is a plugin called "PluralEyes" from Singular Software which lines up video clips by their audio. You could record the audio on a separate recorder and use PluralEyes to line them up using the lower quality camera audio as a sync guide.

Singular Software

March 6th, 2010, 07:50 PM	#17
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Yeah, I'm that guy. :) Overall, the juilcedLink into the 5D2 is surprisingly good. The H4n is nearly as good. The Microtrack II is not so good - unless you drive it from external preamps, and even then it has audible flaws. Overall, the JL/5D2/ML combo is VERY clean, and amazingly crisp, but slightly harsh. The H4n is about 6dB noisier and slightly soft or mushy by comparison. Both give professional, but not audiophile, quality. juicedLink now makes the DN101 for fooling the auto-gain. It also has a headphone output and meters. I haven't had a chance to test it yet. So, why is the JL/5D2/ML slightly harsh? I believe that I'm hearing some high frequency aliasing. When Canon updates the 5D2 from 44.1 to 48 kHz, this could help. Of course, the other Canon cams already do 48 kHz. Why is the H4n a bit mushy? I believe that I'm hearing phase distortion. If the highs and lows are slightly delayed from one another, the result is a smeared edge. This isn't something that is fixed by simply cranking the HFs. Similarly, the JL/5D2/ML isn't crisp because the highs are boosted. The frequency range sounds pretty flat to me. But the differences are slight enough not to worry. Both solutions will sound great for many applications. I've heard our JL/5D2/ML recordings in a theater (48 Hour Film Project), and the noise level was nice and low without any noise reduction being applied. Still, it was slightly harsh in that environment, but our sound quality was among the best of those who participated. I'd want better for a Hollywood release though... __________________ Jon Fairhurst

March 6th, 2010, 09:26 PM	#19
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Yeah, it should sound good, due to the clean preamp. I haven't tested it though, and I don't own a 7D, so I can't offer any details. __________________ Jon Fairhurst

March 7th, 2010, 01:36 AM	#22
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	Again, I don't have a 7D and I haven't tested the DN101. *[NOTE: I've edited the following for accuracy.]* I do have the 5D2 and have tested it with the juicedLink with Magic Lantern. I set the gain to +17 dB in analog and zero in digital, preserving the maximum dynamic range. With my specific test setup, noise hovered around -96 dB, which would be the theoretical limit of the 16-bit recording. The results were about 6dB better than the H4n. And I don't know why the juicedLink would be a "funky" preamp. Okay, the name is kind of funky :), but this thing has some of the cleanest gain that you'll find. (I see you were talking about the Canon preamps, which aren't super clean on their own, but exhibit no bad behavior. Note that the Canon AGC is done digitally, not in the preamp.) From my tests with the Beachtek and 5D2 with stock firmware, the signal jamming approach drives the digital gain down to or near zero. The problem was the analog section. It's set on the 5D2 to a fixed +31 dB. It's possible that even with a clean signal, the noise could be around 14 dB higher than on the 5D2/ML setup. It could be higher, due to bleed from the AGC defeat signal, but it really needs to be tested to know for sure. It will be interesting to see how Canon programmed the analog gain in the new 5D2 firmware. They have one adjustment. Does it control the analog side, the digital side, or does it mix the two? My feeling is that with the expected noise, some will be happy with that performance with the advantage of not having to sync. Others will prefer a double system. A third option is to use a field mixer into a top recorder and to feed the line level into a juicedLink and 7D as a clean sync track. This is especially good if you are using a wireless system and the 7D internal mic wouldn't be able to pick up the dialog. It also gives you a backup track, if something goes wrong with the primary recording. __________________ Jon Fairhurst Last edited by Jon Fairhurst; March 7th, 2010 at 01:42 PM.

March 7th, 2010, 01:35 PM	#28
Jon Fairhurst Inner Circle Join Date: May 2006 Location: Camas, WA, USA Posts: 5,513	@Brian: I haven't had a chance to try the DR-100, but would love to. It gets consistently good reviews. @Marco: You're right. I didn't do these tests with calibrated lab equipment. I did the test relatively. So I shouldn't claim absolute numbers for the units. My mistake. But I did do the tests consistently from unit to unit. The nice thing about the dB scale is that it's relative. I first measured the sensitivity of each unit with a 500 Hz tone. I wasn't able to simply calibrate everything, specifically, because the Beachtek has no gain and it defeats the Camera AGC, so there's no gain adjustment there either. Similarly, the Microtrack II has no analog gain adjustment. The H4n and the JL/5D2/ML were the only units where I could boost gain in the analog domain. My assumption is that if I used each of these systems with the same mic and performance, that I'd have to boost or lower the gain in post to achieve the desired mix. So I applied gain to my noise recordings based on the calibration levels that I had measured earlier. The details of the noise test are here: 5. Canon 5D Mark II Audio Exposed - Noise on Vimeo And you can download an uncompressed recording here: http://p3pictures.com/audio_5d2/Audi...art5_Noise.zip Regardless of the numbers, the juicedLink clearly has the lowest noise - especially as we move up the frequency range. Frankly, the thing is eerily quiet, considering what I had expected plugging into a 1/8" jack on a still camera. Watching the video again, I had forgotten that I ran the analog gain of the camera at +17 dB. (I've been running +10 dB these days.) So, with the 7D analog gain likely fixed at the maximum of +31 dB, I'd predict that the DN101 will have at least 14 dB more noise than my JL/5D2/ML test produced. It could be more, due to the bleed of the AGC jamming signal. But again, I haven't tested it. And thanks for reminding me that my tests were not absolute, but relative. I shouldn't have quoted the numbers as such. __________________ Jon Fairhurst

March 6th, 2010, 08:25 PM	#18
Corey Benoit Major Player Join Date: Feb 2010 Location: www.faymusmedia.com Posts: 203	so a dn101 and a juicedlink cx231 combo with a studio recording condenser mic on a boom pole will sound good right? will it blend with my ex1r? i will be recording from boompole into that camcorder and using the 7d for alternate scenes...

March 7th, 2010, 12:32 AM	#20
Marco Leavitt Inner Circle Join Date: Jul 2002 Location: Albany, NY 12210 Posts: 2,652	For those who are happy with the results you're getting, more power to you. For me, I would never trust a setup that relied on fooling the autogain of a camera with funky preamps for anything critical. Can you really be sure that you are always calibrated correctly? To say that you heard sound recorded to the camera in a theater and it sounded fine, well, I'm not sure that really tells us anything. If you record a nice strong signal to practically any device it can sound all right. It's what you don't know you are missing -- that's where the real loss is. But I would say, definitely, that if you concentrate on good recording techniques, with good mic placement, proper levels and gain staging, you can get surprisingly good results with the humblest of tools. So I'm not saying that audio recorded with the Juicedlink/7D combo is necessarily going to be bad. Just for me personally, it all just seems a little too wonky to trust. It would be really great if somebody were to actually conduct a test of the signal to noise ratio, frequency response and distortion of this combo. It would sure take a lot of the subjectivity out of the debate. There certainly do seem to be a lot of people who would like to record direct to that camera. Also, even with the cleaner preamps of the Juicedlink, you can't bypass the preamps in the 7D altogether. All you can do is force them down, and at a certain point you are increasing the distortion and possibly encountering an impedance mismatch. The sweet spot on the gain for most camcorders is usually about one third of the way up from the bottom. I don't know how you can be certain exactly what the internal preamps are doing on the 7D.

March 7th, 2010, 01:20 AM	#21
Corey Benoit Major Player Join Date: Feb 2010 Location: www.faymusmedia.com Posts: 203	recording directly to camera saves time, and is way more practical and convenient with todays technology... the only reason that the film inustry recorded seperately was to get better sound qualtiy and didnt have good recording directly with the cameras... todays cameras have the ability with technology to record high quality sound...

March 7th, 2010, 05:03 AM	#23
Brian Luce Inner Circle Join Date: Jan 2006 Posts: 2,290	Jon have you had a chance to play/compare with the Tascam DR100? I exchanged my Zoom for the Tascam because I liked to features better, never tested the audio fidelity though.

March 7th, 2010, 05:32 AM	#24
Corey Benoit Major Player Join Date: Feb 2010 Location: www.faymusmedia.com Posts: 203	will this work with the 7d? it looks like it will.... http://www.bhphotovideo.com/c/used/531903/Cavision_RS1580S_SPE_RS1580S_SPE_Shoulder_Pad_with.html

March 7th, 2010, 11:17 AM	#26
Corey Benoit Major Player Join Date: Feb 2010 Location: www.faymusmedia.com Posts: 203	also what is the best way to edit 7d footage in premiere pro cs4? i have avch and h.264 codecs and presets....the 7d records in .mov files, and its really slow... heres my computer specs: core i7 920 3.4ghz 8 mb cache triple channel memory 3 x 1 gb ddr 3 @ 1.7 ghz SLI 2x geforce 8800GT's both overclocked WD hard drive 32mb cache do i need more ram, my cards are pretty fast...

March 7th, 2010, 11:32 AM	#27
Marco Leavitt Inner Circle Join Date: Jul 2002 Location: Albany, NY 12210 Posts: 2,652	Jon, when I referred to funky preamps I was talking about the 7D, not the Juicedlink, which by all accounts has very nice preamps for the cost. Regarding your tests, and I mean this respectfully, but I'm skeptical about your test results. If you are really getting 96 dB SNR with that system, then you are achieving better results than any camcorder, or even any 16 bit recording device that I'm aware of. In fact, I don't think that kind of performance is even possible on a 16 bit recorder. I think you need to review your workflow and software, and maybe test it on another device with published specs to see if you can replicate their results. I do very much respect that you are trying to go about this in a scientific way, so please don't feel that I am being combative. Even 75 dB SNR is quite good, and better than many professional camcorders with actual XLR line inputs.

March 9th, 2010, 11:28 PM	#30
Laurence Kingston Major Player Join Date: Aug 2004 Location: Clermont, FL. Posts: 941	Here's a completely different way to do it if you are working with Sony Vegas or FCP. There is a plugin called "PluralEyes" from Singular Software which lines up video clips by their audio. You could record the audio on a separate recorder and use PluralEyes to line them up using the lower quality camera audio as a sync guide. Singular Software