Update 2013-04-18: The impedance for audio input on the GH2 is apparently 4.7kOm. Thanks for @aenima for testing that!
In the 30 minutes or so I spent searching for others' information on the audio hacks, I didn't find much so I'm starting a new thread.
I'm still parsing through data from 12 hours of testing the hacks myself, but I wanted to invite everyone else to post their data too. I'll be posting ideal settings for using the hack to set levels appropriate for interviewing using the onboard mic or the 3 gain settings on a Rode Video Mic Pro shortly with plans to update with settings for another Sony lavalier mic within a week or so.
All the best, Per Lichtman VP/Music and Creative, Beyond Belief Music Corporation Founder of Pasadena Pulse
I tried some of the hack's higher bitrate audio settings and found that the improvement to audio quality were limited, and they removed the ability to play back clips in-camera. However, I mostly record double-system sound anyway.
Glad to see some participation already. Here are some of my thoughts and questions.
1)Can anyone else confirm the range of gain adjustment, both with factory and hacked settings? In my own tests, the factory AGC settings allowed an adjustment range ca. 15 dB, though Level 4 gain was difficult to measure due to very aggressive compensation. The range maximum adjustment range I found early testing was ca. 50 dB, making the 20 dB gain boost on a Rode VMP much useful.
2) The tooltips had indicated that the final two hexadecimal digits served a different purpose than the first two, but did not elabore greatly on either. My early testing showed that the first 2 digits controlled the aggressiveness of the AGC and the limiter, with the value 00 resulting in the least of either so far, while FF resulted in very aggressive processing. As could easily be inferred from the factory values, the last two hexadecimal digits seemed to more directly control overall gain level, with lower values providing ,ore gain than high ones.
However, in my tes recordings with gain values, the response did not seem entirely consistent throughout the 256 values represented by those last 2 digits. Thus, I wanted to inquire if others were finding the same or not.
If the rest of test recordings proved consistent with the early ones, then users would get the best audio quality by always leaving the first two hexadecimal digits at 00 and setting the last two to cover a wider range than the factory settings instead of dealing with audible swelling and pumping potential.
However, I was curious as to whether 00xx hex values are just less aggressive in processing, or whether they disable AGC artifacts entirely. My first recordings were too few to be sure, so that's why I'm getting ready to review the rest and would really like others' findings as well.
All the best, Per Lichtman VP/Music and Creative, Beyond Belief Music Corporation Founder of Pasadena Pulse
I'm using a pair of Panasonic WM06t in a Sony body (threw the Sony electrets out .. rubbish) The Panasonic are Back Electret and are very flat and no peaks 20Hz to 20Khz there are Omni so will not suit all applications.
The GH2 self powers the MIC capsules
I do Video's of musician friends at GIG's
It would be good to have a setting in MIC Level for a particular microphone OR each of the 4 level settings able to handle and adjust for the SPL of muscian bands for a particular Microphone while minimizing compression and limiting.
Excellent research thepalalias. I am also eager to find out more about the ability to suppress the agc as much as possible. I am aware of two stage recording for audio, but would like the option to do good quality recordings without a raised noise floor when shooting a scene without dialog while using the external mic input.
It would be useful to force the audio to stay in external mic input mode and ignore the onboard mics when the plug is unplugged.
The reason being that when the cable is accidently unplugged, you won't be fooled into thinking you are recording audio from your external input because the onboard mics are making the levels dance anyway.
I know there is a little microphone icon that I can look for, and will be watching it more often, but I'd rather keep my focus on the audio levels instead.
For now I'm considering gluing my adapter to the camera :)
Wish I had more time to post: it's been an insanely busy week. But here's a few findings from the field so far.
1) Modifying the gain settings seems to work very well with the internal mic and 2 (out of the 3) settings on the Rode Video Mic Pro. However, there have been issues using the +20 dB setting. Namely, even though it's easy to get the digital level to stop going into the red, distortion seems to be occurring at an earlier stage. So if you send a signal through with even moderate SPL levels (such as someone playing acoustic guitar and singing from 8 feet away) it both distorts AND comes through with low level. It's possible that using an attenuator cable may help.
2) As a result, I've mainly been using the +0dB gain setting on the Rode Video Mic Pro and tweaking the hex in the GH2 hack.
SuperRoach: So far, 00xx (with xx being numerically high values, thus providing low gain) has been giving a pretty consistent noise floor without noticeable fluctuation. It's never going to sound as clean as the $3,000-$13,000 signal paths I use when recording Joanna St. Claire, but I don't think any of us are really trying to get there either. If you end up needing that and go the second system route, the cleanest portable signal paths you can get are hooking up Sennheiser/Neumann digital mic (I use the KM 184D with the battery pack option and SPDIF coaxial starter kit) with the built in converters into a digital recorder with a digital input (like the discontinued MicroTrack II). That gives you a signal path that I had a Grammy winning classical vocalist prefer over her normal boutique pre-amp and analog Neumann and Rode Classic. Was around $2,500 at the time for the whole package (including cards for the recorder) so you should be able to pick it up for much less now.
All the best, Per Lichtman VP/Music and Creative, Beyond Belief Music Corporation Founder of Pasadena Pulse
@thepalalias thanks for sharing. I'm also using the Video Mic Pro, but struggeling to find the right audio level without tweaking the hack. Could you give me a hint of what numerical values you use combined with Video Mic Pro (@ +0dB gain). In one (or several) circumstance(s)? thanks again
Hi, This is very interesting for me, as I plan on using lavalier mics for interviews. I don't know much about hex settings, but please tell us your experiences tweaking the various audio settings in PTool. Thanks Izhar
Currently, the Rode VMP settings I've been using are four the values on the right below. They are listed from quietest to loudest and in the decimal form they were input, not the hexadecimal, along with original factory values on the left.
None of these values produce noticeable pumping artifacts and all but the new AGC 3 apply less in-camera gain than the quietest factory value.
I'm still working out the best value for AGC 0, since I'd originally set it to try to deal with the high input level of the +20dB setting on the VMP, but that's proved ineffective. Nonetheless, I keep the 235 in there for any potentially really loud scenes, but don't really use it as yet. The majority of the time I have it set to either 203 or 36 if the gain is set to +0dB gain, as you specified. Example applications (taken from the homeless documentary I'm shooting) would be 203 for recording a nearby guitarist and vocalist and 36 for recording a quieter interview in a low noise setting. Hope that helps as a starting point.
I outlined some of my experiences in my post to @evero but here are some additional thoughts.
- While I'm still optimizing my settings for different uses and microphones, I will say that the factory setting for AGC 3 is (in my opinion) something that should not be used if you care about audio quality. Replacing it with 14 gives you a lot of gain without the pumping artifacts.
- Different microphones will record at different levels, but the good thing about the 4 values (14, 36, 203, 235 in decimal not hex) is that they basically cover the entire gain range the GH2 can provide. You can't fine-tune precisely, but you can get in the ballpark with pretty much any mic you throw onto it.
- General recording advice: when you're setting your level using the onboard level meters with the values above, be more careful about "too loud" than "too quiet". Pick the setting that lets you see activity on the meters, but pick the next quieter one if you see them go into the red. This has worked well so far on the interviews (with the usual caveats you would expect from the environments in which you would conduct interviews with homeless individuals).
But, the "low number" actually gives you the highest gain. So when using an external pre-amp, you would likely want a value in the 203 to 235 range as your lowest value. 235 is one of the quietest useful values I've been able to use so far.
It's not the noise that's the main issue, it's the distortion. If you use the AGC 3 value I specified earlier with the VMP gain set to +20dB, you will often find situations where you have the following:
- Level meters show audio signal as fine. - The final waveform you get into your computer is nowhere near clipping - tons of gain level headroom left but... - ... you are still getting tons of distortion on even moderately loud input signals.
Weeks ago I had several takes recording a homeless man I was interviewing playing guitar that were unusable because I didn't find out about that issue until I got the files on my computer. Live and learn.
So if you are going to use the +20dB setting on your GH2, the only avenue I have not tried yet is using an attenuator cable to reduce the signal by ca. 10 dB (leaving your input at +10dB instead of +20dB). In the meantime, I would strongly discourage using the VMP +20dB setting with the GH2 unless someone else has found another workaround.
What's about an hack that enables to capture 3 audio lines? R, L for stereo (onboard mics) and additional the mic plug for an external microphone? Or maybe stereo sound summed on one channel and the external mono mic on the other channel?
@anyone, it would be nice to be able to patch the GH2's audio so that it will further less compress the Audio. Since HD Audio on Blu-Ray discs is far higher quality. On the PS3 it displays Audio bit rate @ something like 1 Megabyte per second or more. I could be reading it wrong and it actually be mega bit.
The short answer is that 1,500 kilobits per second is a compressed datarate that has been available on DVD for years, but you actually listened to audio at 1,411 kbits uncompressed for years before that, whether you knew it or not. Blu-Ray audio CAN be delivered in much higher fidelity, but it often is only somewhat higher quality. Suffice to say, if you're only getting 1,500 kbps, it's not really HD audio (or rather, it's the absolute minimum that could get away with the term).
I wrote a really long post in response (roughly twice as long but I saved it somewhere else) but the short version is that TrueHD uses Meridian Lossless Packing (MLP), with lossless being the operative word. The version of the specification used in Blu-Ray maxes out at ca 18,000 kbits. That bitrate supports up to 24-bit, 96 KHz for for 7.1 or 24-bit, 192 KHz for 5.1 thanks to the compression. The typical compression ratio is somewhere around 1.5:1 (lossless), vs a minimum compression ratio of 3:1 (lossy) with the maximum datarate on the GH2 hack.
With 1.5:1 ratio, the most you could fit into 1,500 kbps is 2 channels of 48 KHz, 24-bit, with no options for surround sound. And 48-24 is the minimum format sold as "HD" to the audiophile market, the minimum I would willingly record music I was going to engineer in a studio (and I always tell my students to make sure the recording equipment they buy supports 96 KHz, 24-bit, but that's it's own tangent).
DTS-HD formats also support a similarly impressive array of high resolution formats, generally lossless (I wouldn't promise "all" without double-checking the compatibility options that got built into every single variant).
Returning to the GH2, the codec itself is meant to be used for "lossy" compression, not "lossless" and Vitaliy would essentially have to write code to swap in another codec to get either lossless compression or access to higher bit-depth or samplerates. Even if we did, we would be limited by the quality of the analog to digital converters in the GH2 and the seeming inability to bypass the microphone pre-amp completely. In other words, all sorts of issues that create problems before the audio even gets compressed. Those are the ones I've been trying to find the best settings to address. :)
If we can't have true lossless compression or high samplerates or bit-depths, etc., etc., I would have to say that there is little be gained by going past 448 kbps for stereo recording (and that's coming from someone with experience consistently picking out specific analog to digital and digital to analog converters in blind testing). There's just a point of diminishing return and nothing short of full lossless gets beyond that. If you want the highest fidelity, there are other options available.
But, the hack has given us way more control so that we can do a better job of recording interviews, dialog, etc. when we want "the reasonable best" in situations where using additional equipment isn't practical or cost-effective. That's a really good thing and I'm enjoying it.
@thepalalias, thanks for the info it seems you know your stuff and like me who did a B.S.c Creative Music & Sound Technology. Yes, like some other problems with the hack like spanning, it would require things like a complete re-write of the whole operating system. As you told me and I told someone else who commented on one my videos on youtube- Even though its not perfect, its far more cost effective than buying the broadcasting stuff.
@chrimsbroome A great area to get your degree in. Mine was in Music Theory/Composition but I had already founded a music label and started doing consulting work for music tech. companies at the time, so sounds like we might be on the same wavelength. :)
It's not just a question of the money, too: it's the time, effort, manpower and extra equipment (and things to keep track of). If you're shooting by yourself, being able to shoot audio on camera at the same time by yourself is a lot more practical than having a second recorder. If you have someone around that can handle audio at the same time you're shooting, then that's a different story. :)
So you would say that for this kind of job (interview, atmo, dialog) the quality of the inbuilt preamp is suitable? Sure, not perfect at all, just good enough... ? Does it make sense to add a small external preamp like Tascam iXz, so you can keep gain in cam as low as possible (and maybe get xlr in as well)?