GH2 Why do we need a GH3? Isn't the GH2 enough? What should Panasonic do for GH3?

[questech]

Hello TheDingo, thanks for the feedback!

I know that you are using a C-mount lens and have to use ETC mode to get proper coverage of the sensor, but I think this is HUGE mistake, as ETC mode cripples the low-light and low-noise ability of the GH2 sensor. This is why I have never bothered with ETC mode and also why I won't go near C-mount lenses.

Yes, more noise for sure, but I haven't noticed any exposure penalty. The lens is actually pretty cool and is a good tool for certain shots. I also have the Lumix 14mm - 140mm as well as several Nikon lenses, both zooms and primes. Using the C-Mount lens actually triggered my quest for answers because ETC Mode does induce much more noise.

My experience has been the exact the opposite of what you are describing.

I've been using the Driftwood Quantum 100 ( 100 Mbps I-Frame ) patch for a few months now ( I've shot 2 Indie films, a dozen 10-30 minute interviews, a classical concert, a sports infomercial, and another Indie film is in the pipe to shoot in about 6 days ), and I am as happy as can be with my GH2. I get beautiful, sharp, low noise images ( 160 - 800 ISO ) using SanDisk 64 GB SDXC UHS-I 95 MB/sec memory cards that enable me to shoot up to 84 minute takes. Using the Quantum 100 patch and these memory cards my GH2 works flawlessly. First time, every time.

That's very cool, I'm glad you are having such good experience with it!

What I don't see (from what I've looked at) is where's the big improvement?

The footage I have watched has had all kinds of subtle imperfections, and I just can't see how it's any better than the stock firmware. And the added expense for memory cards isn't that much, as you pointed out, but the point I have been trying to make all along is; if the stock firmware produces images just as good or even real close to the hacked firmware, why take the storage space hit??

Now I already know you disagree with my last statement, and I am not saying you are wrong by any means. I am just saying that I haven't seen the improvement that everyone is talking about.

However, with all of the people that share your opinion, I can't help but to think that I am missing something... I'm just trying to figure out what that is!

I do agree that image noise becomes a problem once you go past 800 ISO ( I will use 1600 ISO if I have absolutely no other options ), which can be cleaned up with NeatVideo to some degree, but if it's a paying job and I need affordable low-light performance then I would rent a Sony HS-100/HS-700 camera. ( if the GH3 doesn't pan out this year, then I will look at buying a HS-700 body )

I own a production company we use high-end gear for paying gigs, this GH2 is for my personal use and to make YouTube videos, but my next personal camera is going to either be the Black Magic Camera (my partner is lobbying for that one) or the FS100 or preferably the FS700, so I'm right with you on that one!

Yes, the SanDisk cards are expensive, but with just two 64 GB cards, I can continuously shoot and dump to a laptop all day.

I used a 64Gb SanDisk Extreme 45MB/s for my hack tests and I still got an error (not fast enough) with a couple of the patches, and I only shot stationary scenes.

...I'm sorry Questech, but you are going to have to illustrate this opinion with some direct examples, as I really have no idea what you are talking about. ( I'm a professional studio photographer with 20+ years experience, and have been shooting video for 10+ years, so I have a pretty good eye too )

Well, I'm probably a little older than you, but I too have been in business for a while. I started out as a photographer and in 1980 I started my first production company. So let's both agree that we have good eyes :huh:

I will find the footage in question and provide links. But the problem that I have with the footage I viewed, is that it must have been compressed at least once, and the imperfections that I'm seeing could be related to that. I did download some footage from Vimeo that eluded to the fact that it was original footage, but GH2s don't put out AVIs or MOVs, so that couldn't have been true.

I would love to see some actual footage right out of the camera that shows this quality improvement, because so far there's just no evidence of it that I could actually see with my own two eyes that supports that. And don't take that the wrong way, I just haven't seen the results you and many others are referring to

I have seen my tests and my unhacked firmware produced a cleaner picture, with less noise than any of the patches I tested.

BTW because I was only concerned with quality in ETC Mode at the time that I ran my tests, all of my tests were in ETC Mode.

I certainly wouldn't hold this position if I hadn't spent hours investigating, after all, we all have businesses to run and I'm up against some deadlines as we speak.

I would love to become a convert. If you have access to, or know where I can download some MTS files of high quality hacked footage, I'd be grateful... even better hacked and the same exact scene unhacked.

Thanks!

 

[Per Lichtman]

There seems to be a difference of opinion here, but I think that Anthony's underlying point that it would be great (if possible) if the GH2 could improve the noise performance in EX modes is well taken. If someone has to primarily use EX/ETC mode on the GH2, they are likely to have a lower opinion of the noise performance than someone that is primarily relying on the downscaled output that uses the whole sensor.

In my experience with the standard downscaling, the noise performance is competitive for the price class when working properly - but there are some very annoying things that can happen under isolated circumstances where shadows will start to flicker, etc.

In either case, I think that improving noise performance in both modes can only benefit the user. My only caveat is that I think it is important to cater to the part of the community that wants to do the reduction in post just as much as to the community that wants the reduction in-camera.

For me, the biggest thing I want is more consistent performance - fewer flickering shadows, no dual horizontal line bug becoming more obvious at high ISOs, etc.

I don't see how anyone loses if the situation improves.

As for my own read on the camera comparisons: I've shot RED, spent some hands-on time with the C500 and shot for over a year with GH2 and GH1. The GH2 is a great camera for the money and has great image quality. But there are many areas where it cannot begin to touch the RED.

The Zacuto test is by many accounts a very good example of what they set out to test, but what people are interpreting it as seems to be something different. A strength of the RED is dynamic range and the post-processing file format - an area where the GH2 cannot compete with it. The part of the test that most people are referencing is the one where people where able to modify the lighting and ND filter etc. within certain parameters. It is also a section in which the grading was done according to individual preferences. So it does not say very much about the performance capabilities of the cameras - it says something about the potential application of the tool within that workflow.

If you don't need RAW, if you don't need the resolution or frame rates and if you don't need the dynamic range, or uncompressed audio support, then you won't be able to appreciate the advantages a RED camera can have over the GH2. But on the flip side of the coin, the GH2 is a lot cheaper (both the camera itself and the accessories like batteries and storage), smaller and has a fully articulated LCD and an EVF included by default - it is also a lot quieter to use.

If you nail your exposure and settings on-site with the GH2 and can control the dynamic range of the scene as well as shoot at low ISO (without the ISO bug) and use similar lenses, you can get very similar footage with the GH2 and RED. But you really notice the difference when you start trying to push that footage in a different direction in post, or to bring up shadow detail or recover highlights, etc.

 

[Per Lichtman]

I have some MTS files you can download but they are not a great example of what we are talking about. They are shot in 24H, not EX24H.

http://perlichtman.com/pasdenapulse...2-new-patch-tests-sedna-a-b-c-released-today/

Here are the settings tested.

4B Orion, 4D DM1, 4D DM2, 4D DM3, ClusterV1, Factory 1_11 Unlocked, Sedna AQ1 A. Sedna AQ1 B, Sedna AQ1 C, Sedna Q20 A, Sedna Q20 B, Sedna Q20 C.

Also, for stability the card you want is the SanDisk 64GB 95MB/s. The performance delta between it and even a Delkin 16GB 95MB/s or a SanDisk 64GB 95MB/s is important for spanning on the GH2 and I have demonstrated a difference in the field. You can see some of the test results here.

http://perlichtman.com/pasdenapulse/wordpress/gh2-hack-reference/

 

[questech]

I have some MTS files you can download but they are not a great example of what we are talking about. They are shot in 24H, not EX24H.

http://perlichtman.com/pasdenapulse...2-new-patch-tests-sedna-a-b-c-released-today/

Here are the settings tested.

4B Orion, 4D DM1, 4D DM2, 4D DM3, ClusterV1, Factory 1_11 Unlocked, Sedna AQ1 A. Sedna AQ1 B, Sedna AQ1 C, Sedna Q20 A, Sedna Q20 B, Sedna Q20 C.

Also, for stability the card you want is the SanDisk 64GB 95MB/s. The performance delta between it and even a Delkin 16GB 95MB/s or a SanDisk 64GB 95MB/s is important for spanning on the GH2 and I have demonstrated a difference in the field. You can see some of the test results here.

http://perlichtman.com/pasdenapulse/wordpress/gh2-hack-reference/

You are right on as usual. Your previous post hit the nail right on the head!

Thanks for the links, I am off to an appointment but will view them later.

Thanks Again!!

 

[Per Lichtman]

Thanks - I hope they are of some small use to you. I mainly notice the difference in sharpness in things like the text.

Here is the compressed version of a test made using some of that footage.

Password: Driftwood1

 

[Ian-T]

Thanks - I hope they are of some small use to you. I mainly notice the difference in sharpness in things like the text.

Here is the compressed version of a test made using some of that footage.

Password: Driftwood1

Ohhh....I'll play...

Let's see....from 1-5 seconds the hack is obviously on the left. Then at 5 seconds in you flip it and I can tell the hack is on the right hand side.

At ten seconds in the hack is obviously on the left hand side and the stock is on the right. At fifteen seconds you switch it up and the hack is on the right.

How in the world some folks say they don't see any advantage with the hack is beyond me. This is just one small area where the benefits are obvious.

There's something about the stock firmware that smooths out the image (taking away the resolution)....especially noticeable in motion. The GH-1 suffered this and so did the GH-2. The Hack(s) allowed more color AND noise to come through. But my Neat Video loves the footage because it can now better discern REAL noise. I just hated how it handled the noise/compression artifacts before.

Also, I seem to not want to remove the noise lately because it looks different....finer...I can't put my finger on it.

 

[TheDingo]

What I don't see (from what I've looked at) is where's the big improvement?

The Quantum 100 I-Frame Only patch that I've been using does two main things for me...

1- Removes Long-GOP compression effects from motion, so you end up with smoother looking motion ( especially if you shoot 24P at 1/40th sec, or 25P at 1/50th sec )

2- By removing Long-GOP compression you also get rid of most macro-blocking, so fine details are better, and the image grain becomes more like a uniform fine powder than compressed mush ( image sharpening works better when Long-GOP is gone )

If the stock firmware produces images just as good or even real close to the hacked firmware, why take the storage space hit?

It's a choice each of us have to make. For myself, I don't mind the additional size from the 100 Mbps patch, though I haven't been convinced of the need to use higher data-rates.

but my next personal camera is going to either be the Black Magic Camera (my partner is lobbying for that one) or the FS100 or preferably the FS700, so I'm right with you on that one!

I am going to wait for the 2nd generation Black Magic camera, as I have some issue with using EF mount lenses with such a small sensor. If a Micro 4/3 version becomes available, I will probably change my tune.

I used a 64Gb SanDisk Extreme 45MB/s for my hack tests and I still got an error (not fast enough) with a couple of the patches, and I only shot stationary scenes.

Yes, you have to use the 95MB/sec cards. I would probably give up on the high-bit-rate patches if I could not shoot with reliable spanning. Eventually there will be less expensive cards that can handle these data-rates. I am hoping that the Patriot EP UHS-I cards will be the first to do this.

Well, I'm probably a little older than you, but I too have been in business for a while. I started out as a photographer and in 1980 I started my first production company. So let's both agree that we have good eyes

I went to high-school in the 70's, so I doubt that I'm too far behind. ( my start with commercial photography was working at an advertising photo studio in London, England in the early 80's, mostly shooting car accounts and life-style studio sets 6 days a week, 12+ hour days, which was fun but sometimes very challenging. The longest car shoot was 76 hours straight through non-stop where the only men left standing was one other photographer and me, we were out of our minds at this point... )

I will find the footage in question and provide links. But the problem that I have with the footage I viewed, is that it must have been compressed at least once, and the imperfections that I'm seeing could be related to that.

Yes, online footage doesn't do the high bit rate footage justice. Here's a frame grab from an Indie film I shot last March using the Quantum 100 patch at 400 ISO. This is straight out of the camera. I was never able to get this type of result using the stock firmware...

GH2 Quantum 100 - 400 ISO Frame Grab

EDIT: I should have posted a PNG, as I can see some JPEG compression in this sample. ( i.e. the skin tone of her neck in shadow is uneven in the JPEG, but it's fine in the PNG ) I had this JPEG on hand as I had to send the producer some samples from the shoot. I don't think I kept the original PNGs used to make these JPEGs. ( the aliasing on the edge of the fake magazine is a GH2 sensor issue )

 

[questech]

Hello Again TheDingo... You Da Man

I guess we are about the same age - with it comes wisdom.

The Frame Grab, even for a jpg is great! I would like to see it next to a non hacked version though. The problem with comparisons, is that there aren't many apple to apple ones. Unless someone sets out to do a test, like I did, they usually are showing footage as a result of a hack... most people aren't going to change firmware in the middle of a shoot just for the fun of it, so we rarely get an apple to apple comparison.

I'm not questioning you here, but for the sake of higher education, maybe the stock firmware may have given you just as good of a frame grab?? But unless you can go back in time and shoot an unhacked version of that Exact scene, we will never Really know.

I have made setting adjustments just changing the GOP to 3 and it didn't have any adverse effects on my noise levels, so eventually I will test it on some moving scenes (all my tests were on a stationary target).

I didn't have any idea that I would be writing about this, so I didn't save any of my tests, I just decided that the hacks weren't going to net me any mileage and forgot about it. I also shot all of the tests in ETC Mode because that's where my concerns were, and the results were very clear, all of the patches that increased bitrate worsened the noise, period... no question about it.

Now here's another annoyance when people discuss noise, I always see a single frame (no slam on your frame grab!) and that's not a complete or definitive way to test for noise. Noise in photography is way different than noise in video in that video is way less forgiving. What's acceptable in a single frame can become a nightmare at 24fps. Areas that look smooth in a single frame become moving masses of "blobucals". Some of that might not be noise, but I think it stems from noise in some way.

And the only way you can test these patches is to look at the Out-of-the-Camera clips, otherwise you have additional compression to deal with. So it's frustrating.

And just as an aside, I can see how my position can be considered controversial, especially when so much time, energy and hope have gone into these patches in an attempt to improve the work we do, but the only reason that I am investing my time in this discourse is that it might help in some way further the discussion, and thereby the cause. Hopefully Panasonic is listening to the end users and our discussions may contribute to better codecs.

I am in deadline right now, but I am installing CS6 and will try to do some testing so that you can see my results and you can see why I am so frustrated.

Thanks for all of your feedback!

 

[TheDingo]

I guess we are about the same age - with it comes wisdom.

...Or the illusion of wisdom. ;-)

The Frame Grab, even for a jpg is great! I would like to see it next to a non hacked version though.

Once I get my blog going ( hopefully in early July ), I will try and put together a "Hacked vs Stock" camera test with movement and a supered magnified window to show the difference in the grain patterns.

I will test it on some moving scenes (all my tests were on a stationary target)

Other than the grain pattern, static images look pretty much the same when comparing Stock firmware vs Patched firmware. You have to "overwhelm" the AVCHD Long-GOP CODEC ( with movement and fine details ) in order to see it start to fall apart, while the I-Frame footage keeps chugging along as if nothing is different.

Noise in photography is way different than noise in video in that video is way less forgiving. What's acceptable in a single frame can become a nightmare at 24fps. Areas that look smooth in a single frame become moving masses of "blobucals". Some of that might not be noise, but I think it stems from noise in some way.

In terms of image "grain", EVERY frame of the footage that I pulled that frame grab from looks the same. This is the beauty of shooting with an I-Frame CODEC, where each frame is a stand-alone "photo" with no dependencies on what came before or after it. This also makes the "grain" completely uniform, no matter what is happening in the image.

but I am installing CS6 and will try to do some testing so that you can see my results and you can see why I am so frustrated.

When you get a chance, please try shooting with the full sensor using Driftwood's Quantum 100 ( 100 Mbps I-Frame ) patch using 24P at 1/40th second, and ISO ratings between 160 - 800. This is the combo that I fell in love with, and have been using since early March. You will have to dig up a high speed memory card in order for this to work, though your SanDisk 45 MB/sec cards might be OK for non-spanning footage.

Look forward to your re-evaluation of things.

Cheers!

 

[Per Lichtman]

In terms of the grain difference, you can already see that in the example footage I posted above.

 

[Lpowell]

What I don't see (from what I've looked at) is where's the big improvement?

The footage I have watched has had all kinds of subtle imperfections, and I just can't see how it's any better than the stock firmware. And the added expense for memory cards isn't that much, as you pointed out, but the point I have been trying to make all along is; if the stock firmware produces images just as good or even real close to the hacked firmware, why take the storage space hit??

During the development of Flow Motion v2, I made numerous comparison shots with the unhacked GH2 firmware. For consumer videos viewed on computer screens, an unhacked GH2 looks very good. But with post-production grading or theater projection, magnification clearly reveals unsightly digitizing artifacts due to its limited 24Mbps peak bitrate. What's worse is that when a scene exceeds its bitrate limits, the unhacked GH2 firmware may silently plunge into Fallback Mode, and produce degraded image quality using very coarse macroblock quantization. In my view, this shortcoming makes the unhacked GH2 completely unreliable for anything but consumer purposes.

For those who're interested to compare for themselves, I've uploaded unedited frame grabs and links to the original GH2 MTS files here:

GH2 100Mbps Flow Motion v2 Patch - Banding & Blocking Be Gone!

 

[questech]

During the development of Flow Motion v2, I made numerous comparison shots with the unhacked GH2 firmware. For consumer videos viewed on computer screens, an unhacked GH2 looks very good. But with post-production grading or theater projection, magnification clearly reveals unsightly digitizing artifacts due to its limited 24Mbps peak bitrate. What's worse is that when a scene exceeds its bitrate limits, the unhacked GH2 firmware may silently plunge into Fallback Mode, and produce degraded image quality using very coarse macroblock quantization. In my view, this shortcoming makes the unhacked GH2 completely unreliable for anything but consumer purposes.

For those who're interested to compare for themselves, I've uploaded unedited frame grabs and links to the original GH2 MTS files here:

GH2 100Mbps Flow Motion v2 Patch - Banding & Blocking Be Gone!

Hello lpowell,

Thanks for producing that, I am going to download it shortly.

I like what I'm hearing. All of my tests were done on a low lit, high contrast scene and my results showed the unhacked firmware to be the winner. Now I was testing for a very specific result, low light in ETC mode.

BTW out of all of the patches I tested, FlowMotion performed the very best for me (followed by Sanity)!

I disgarded all of the footage from my original test because I didn't know that I was going to be writing about the results until I read this thread and one other on another blog.

BUT because of all of the dialog since then, I reshot the tests of my unhacked firmware and the hacked firmware using various patches.

This time I shot the same scene in ETC mode as before, and in nonETC mode using the same lighting, framing and the same exposure.

When viewing the results I used a 24" computer monitor and I watched the footage from 3', 2', and less than a foot to reveal as much detail as possible without adding artifacts due to enlarging the image (I also inspected the clips within PP CS6 at enlarged sizes to crosscheck my results).

I shot both HBR mode and 24H mode and ended up with about 30 15 sec clips. I think the results are revealing.

BTW I shot HBR & 24H for 2 reasons, 1) some of the patches only affected one or the other, 2) I found more noise in the 24H footage in my original test.

Now the question is, where can I upload the HBR footage (14 files averaging 70Mb each) so that everyone can see the results... any suggestions?

Thanks!

 

[Lpowell]

@questech - I'd suggest opening a Vimeo account, where you can upload your original GH2 footage and make it available for download by others.

 

[TheDingo]

Now the question is, where can I upload the HBR footage (14 files averaging 70Mb each) so that everyone can see the results... any suggestions?

Open a free Gmail account that comes with 5GB of storage, and then use the Documents section to upload and share files. Just remember to change the sharing from private to something like share with this link, which is what I do for these situations.

You can upgrade your Google Docs storage to 20GB for $5 per year. ( this might be changing as Google migrates to Google Drive )

 

[questech]

Open a free Gmail account that comes with 5GB of storage, and then use the Documents section to upload and share files. Just remember to change the sharing from private to something like share with this link, which is what I do for these situations.

You can upgrade your Google Docs storage to 20GB for $5 per year. ( this might be changing as Google migrates to Google Drive )

Thanks, I wasn't aware of that!

 

[questech]

@questech - I'd suggest opening a Vimeo account, where you can upload your original GH2 footage and make it available for download by others.

Hello,

I looked into it briefly, but I gathered that the file you uploaded had to be viewable i.e. I would want people to access the MTS file without actually having to watch them. And I don't believe that Vimeo can play MTS files without altering them in some way... is that true?

Thanks!

 

[Lpowell]

@questech - That's correct, Vimeo transcodes each uploaded file into its own streaming format for online viewing purposes. Of course, you don't have to actually watch the video on Vimeo, you can just download the original file. In order to see the original footage, you must download the GH2 MTS file from the video's Vimeo page.

 

[questech]

@questech - That's correct, Vimeo transcodes each uploaded file into its own streaming format for online viewing purposes. Of course, you don't have to actually watch the video on Vimeo, you can just download the original file. In order to see the original footage, you must download the GH2 MTS file from the video's Vimeo page.

OK, so they do what they have to to the original MTS file to get it to play in their player and the original MTS file is available for download from one of the menu items, is that right?

Thanks!

 

[Lpowell]

@questech - Correct.

 

[Per Lichtman]

If memory serves me correctly, only the "Plus" accounts offer the download option on Vimeo.

You could also use DropBox for free (5GB).