GH1 firmware research volunteers required

[Lpowell]

From pure electrical side, if you have short cord from active microphone to even 1/8" mini-jack quality must be the same as balanced XLR.

A short, unbalanced high-impedance audio cable can work flawlessly if you are running your equipment entirely on battery power. But if there are any AC power line fields near the camera, it may pick up noise that a balanced low-impedance cable would reject. In particular, if you're shooting anywhere near AC-powered fluorescent lights or dimmers, I'd use XLR cables on the mics, and drive the GH1 audio input with a low-impedance active pre-amp (e.g. a Zoom H4N used as a mic pre-amp).

[tyampel]

Hello, I took some time to check the remaining languages which are currently listed as unknown and I propose the following changes:

Deutsch --> German
Francais --> French (1)
Unknown 9 --> Thai
Unknown 11 --> Portuguese (1)
Unknown 12 --> Classical Arabic
Unknown 13 --> Modern Arabic
Unknown 14 --> French (2)
Unknown 15 --> Finnish
Unknown 16 --> Danish
Swenden --> Swedish
Portugal --> Portuguese (2)

Portuguese 1 seem to be Brazilian portuguese while portuguese 2 seem to be european portuguese but I cant say for sure right now, the same goes for the two french languages. Perhaps I will give it a try some other time or someone else who speak these languages could lend a hand.

Other than that I just want to say good work and thank you to those involved in this project (especially tester13)

Someone reported that 15 is Dutch - Netherlands/Holland

[Cavemandude]

Ok, I'm still a little confused about this 192kbps AC3 standard/limit on the GH1.

Why is it that the Canon HF200, and I'm sure other AVCHD consumer camcorders, record audio at 256kbps 48kHz stereo AC-3 and not 192kbps?

Here is some good related info also:

http://en.wikipedia.org/wiki/Dolby_Digital

and a screen grab of some specs below

Randy

[Cavemandude]

So they had this 3 (?) camera Canon 5D interview with Cinematographer Shane Hurlbut, Bandito Brother's Jacob Rosenberg, and Hesh Rephun and guess what happened to their audio using what looks like a Zoom H4n? Unusable, they had to use the audio from the camera which they luckily were using shotguns on. Here is what they said happened:

"There’s an audio recorder on a small stand just behind the A camera. Unfortunately it was acting up that day. We ended up using the backup audio from the 5D MKII DSLRs equipped with shotgun mics (with manual audio levels set)."

Imagine if they had not used those shotgun mics directly into the 5D's because they were so confident about how their Zoom recorder was going to capture such awesome audio? Uh sorry Shane we got no audio, can we shoot the interview over again, take 2 ok?

This is exactly why I want to use best possible audio quality in the GH1 and if a higher bitrate is going to provide that then there is no reason not to use it. Scratch audio? I say use backup audio instead on the GH1 just in case things like this happen.

Here is a link to the interview and the audio ain't that great since the shotguns were too far away but I guess it was better than what was on the Zoom recorder and much better than the audio would have been with the built-in mic:

http://www.freshdv.com/2010/05/conve...o-hurlbut.html

Randy

[Ozpeter]

There is no field of human endeavor more fraught with the problems of objective assessment than audio! For instance, see http://drewdaniels.com/audible.pdf where a long term series of tests demonstrated that no individual (in these large-scale tests) could reliably spot the difference between the much vaunted 24/96 audio and normal CD audio, despite the internet being full of people who claim they can. Increasing the bitrate of the GH1 audio is highly unlikely to be evident (and I choose that word carefully) to listeners other than under the most stringent testing conditions, if that. I'd hope that viewers of my material were more interested in the picture and story than in probably unspottable tweaks to audio quality.

But of course if it can be done without slowing up other aspects of the project, there's no harm in individual options and personal choice!

[androoow]

what Ozpeter said +1,........
....although i can tell the difference between say 16/44.1 and 24/96 only certain times the difference stands out ... really to 99% of the world the lower 16/44.1 just works and is why cd's can sound fantastic still after 25 years. I have more interest in the visual aspects being fixed/ improved as these are things that you cant fix/increase beyond the cam , where as there exists plenty of ways to add better audio to a project already, not saying its not important, but to me i come here to see what breakthroughs are happening with the image.

good work to all still..

[aczelkri]

AVCHD audio - why bother?

I, too, have the opinion that the audio part of the AVCHD is not a very important issue. It is pretty good as it is. Just because there is a way of increasing bitrate/sampling rate does not mean that it also makes sense. It is like repainting your car while in fact your engine is broken.

I can't think of any geeks who want to record 96kHz 448kbps studio quality AC3 audio coupled with the muddy picture we currently have. The AVCHD image is what needs fixing. Am I wrong here?

P.S. I am talking about the AVCHD audio here. The MJPEG audio on the other hand does indeed need improvement, as that is audibly worse than the former.

[nubins]

AVCHD compression, what does it affect?

How does the AVCHD Compression patch optimize the settings? Does it change the GOP structure, like adding B frames? What does it optimize?

I just started testing and so far here are the parameters that seem to be working well in 1080 24 (FHD mode).

Video Buffer: 28.000.000
Bitrate Adjustment 1: 20000000
Bitrate Adjustment 1: 24000000
AVCHD Compression
1080p24 GOP Size: 12

Note: As another poster said anything over 20000000 for BA1 and 24000000 for BA2 and the camera locks up as soon as you start recording.

[Cavemandude]

I can't think of any geeks who want to record 96kHz 448kbps studio quality AC3 audio coupled with the muddy picture we currently have. The AVCHD image is what needs fixing. Am I wrong here?

I've never mentioned the desire for 96kHz audio for the GH1, I have no problem with the 48kHz that the GH1 already records at.

So if you just spent a bunch of money on a great voice-over to use in your 30 minute documentary and they only gave you the choice of having it downloaded as a 64, 96, 128, 192 or 320kbps mono audio file, which one would you choose to download and drop into your timeline to edit with? Are you going to just set your Zoom recorder at 192kbps instead of a less compressed setting since most people wouldn't be able to tell difference?

24bit - 96K audio compared to 16bit - 44.1K audio, yes that would be hard for pretty much anyone to tell the difference.

But I think the better comparison is between the Canon 5D which now records audio at 1536kbps and the GH1 at 192kbps. And from what I gather from the Dolby info is that since the AC-3 audio is stereo and using 2 channels, the bitrate is actually 96kbps per channel compared to 768kbps with the 5D. So are you ok with only 96kbps per audio channel to start with before editing?

At least by using Tester13's adjustable bitrate setting at 384kbps you can now get each audio channel back up to 192kbps which I think is more reasonable level. 192kbps (2 channel) stereo might be fine for the final file format quality but I'm talking about the original audio file that still needs to be at the highest quality possible to survive being edited, mixed, EQ'd, remastered and then be re-compressed again into the final distribution format.

Anyways we now have the option to set the audio bitrate (kbps) higher, thanks to Tester13, so it's really no longer an issue since the work has already been done and it looks like it's working.

Randy

[Vitaliy Kiselev]

How does the AVCHD Compression patch optimize the settings? Does it change the GOP structure, like adding B frames? What does it optimize?

Video Buffer: 28.000.000
Bitrate Adjustment 1: 20000000
Bitrate Adjustment 1: 24000000
AVCHD Compression
1080p24 GOP Size: 12

It is pointless set of patches and parameters.
Please see latest Note to Testers, and use it.
For all else read help hints.

[Vitaliy Kiselev]

But I think the better comparison is between the Canon 5D which now records audio at 1536kbps and the GH1 at 192kbps. And from what I gather from the Dolby info is that since the AC-3 audio is stereo and using 2 channels, the bitrate is actually 96kbps per channel compared to 768kbps with the 5D. So are you ok with only 96kbps per audio channel to start with before editing?

This is wrong comparison, you can't compare PCM and AC-3.
Good explanations were already given why.

[Vitaliy Kiselev]

Guys, I need test results made according to latest Note to Testers.

Without them I can't move forward.

[sunsurfh2o]

Sorry, my last post was missing some specifics:

Test was with only 24p enabled, then 28k buffer/24k/20k, then again with same bitrate + AVCHD patch. Shot was static, still objects of varying shapes and colors. Can upload if required but wondering why I'm not seeing any actual change in bitrate stream aside from 'OverallBitRate Max' which I take to mean reported theoretical, not actual during encoding. 'BitRate' and 'Maximum_BitRate' remain suprising par through all corresponding tests. Am I understanding this incorrectly?

[Vitaliy Kiselev]

Sorry, my last post was missing some specifics:

Test was with only 24p enabled, then 28k buffer/24k/20k, then again with same bitrate + AVCHD patch. Shot was static, still objects of varying shapes and colors. Can upload if required but wondering why I'm not seeing any actual change in bitrate stream aside from 'OverallBitRate Max' which I take to mean reported theoretical, not actual during encoding. 'BitRate' and 'Maximum_BitRate' remain suprising par through all corresponding tests. Am I understanding this incorrectly?

Your results are not very useful due to bitrate patches applied, but it seems that AVCHD compression patch worked as expected.
Do not use any other patches other than mentioned in notes!
Read last Notes to Testers carefully (I moved them to first page of this theme!).

P.S. Please, delete your very long mediainfo reports, leave only bitrate related things.

[sunsurfh2o]

1080p:
AVCHD Patch: OBR 7,779 Kbps, BR 7,270 Kbps
No Patch: OBR 15.3 Mbps, BR 14.5 Mbps

720p SH:
AVCHD Patch: OBR 8,260 Kbps, BR 7,732 Kbps
No Patch: OBR 15.1 Mbps, BR 14.3 Mbps

720p H:
AVCHD Patch: OBR 8,277 Kbps, BR 7,748 Kbps
No Patch: OBR 12.4 Mbps, 11.7 Mbps

720p L:
AVCHD Patch: OBR 8,252 Kbps, BR 7,224 Kbps
No Patch: OBR 8,285 Kbps, BR 7,756 Kbps

Better? Forgive my lack of knowledge on this subject, I'm completely new to this technology. Please explain the purpose of your AVCHD patch. What are you hoping to gain by it? I thought the purpose is to increase bitrate, thereby fidelity but the bitrate decreases with the patch and that seems counter intuitive.