Panasonic GH5 Lab Review Update & Clarification

April 16th, 2017
Panasonic GH5 Lab Review Update & Clarification

I published my review of the Panasonic GH5 on March 1st 2017 on the cinema5D platform in order to share the results of my tests about the camera’s performance, with those interested in its video capabilities like myself. Some of the results, in particular, the 10-bit quality in Vlog seemed to be less positive than expected. Many readers commented on my findings and shared valuable feedback. In an effort to re-evaluate the camera’s capabilities based on that feedback I temporarily removed my article from the website and tried to get an opinion on the matter from Panasonic.

After re-evaluation I found that I got different results when using a different conversion codec during my tests. As the Panasonic GH5’s 10-bit files were incompatible with Adobe Premiere Pro, After Effects and Blackmagic Design DaVinci Resolve at the time of my tests, I was forced to first transcode the GH5 10-bit files into another format that could subsequently be used in Adobe Premiere Pro, which is the software I use on a Mac Pro, 2013 for evaluation.

In an effort to make my tests as “real world” as possible, I chose the familiar Apple ProRes HQ codec, not only because it is widely used by professionals in 10-bit environments and external high quality recordings, but also because it was used as the recording format of the Canon C700 that I compared the Panasonic GH5 to and, because it is described by Apple as a codec that preserves “… visual quality at the same high level as Apple ProRes 4444 but for 4:2:2 image sources.” and “… supports full-width, 4:2:2 video sources at 10-bit pixel depths, while remaining visually lossless through many generations of decoding and reencoding.”.

The consequence of using this codec however led to incorrect results when evaluating the Panasonic GH5’s 10-bit Vlog performance as it affected my files in a way that introduced an unexpectedly large quality loss within Adobe Premiere Pro. It seems like the quality issues previously found in my original test were caused by an error of the Adobe Premiere Pro software in connection with the Apple ProRes HQ codec. While the reasons for the error have not been clarified, anybody who intends to use Adobe Premiere Pro in connection with 10-bit files in Vlog recorded on the Panasonic GH5, should consider the conversion steps carefully.

Using the Apple ProRes HQ codec on a daily basis, I was surprised that the conversion affected the Vlog files from the Panasonic GH5 in my initial test so severely. After reevaluation, it has become clear that the Apple ProRes HQ codec is more than capable to hold the 10-bit information Vlog requires and the error in connection with Adobe Premiere Pro needs to be reported and investigated. Furthermore the Panasonic GH5 in Vlog mode exhibits a magenta / green color shift in some parts of the image, that was further exaggerated by said error.

Panasonic told us they have been studying the color shift symptom in V-Log, including the compatibility with each of the color grading apps. And that they will make an announcement on their web site when they release an updated firmware to improve performance.

Despite my best efforts and the intention to bring readers the most honest, unbiased and useful information for real world applications, the choice of codec and a possible software error led to unexpected and incorrect results of my initial 10-bit Vlog test. As a consequence, in order to avoid errors or oversights to affect test results in the future, I have decided to stop doing camera “test lab” evaluations for the time being.

If in the future I will ever do lab tests again these will be done under different test conditions.

Read the updated Panasonic GH5 Review HERE.

Advertisement

40
Leave a Reply

avatar
21 Comment threads
19 Thread replies
0 Followers
 
Most reacted comment
Hottest comment thread
27 Comment authors
Ari KirschenbaumPetter MagnussonGuy McLoughlinAdam Muro Recent comment authors
1
1
- 3
Filter:
all
Sort by:
most voted
 Kevin Almodovar
Kevin Almodovar
MemberApril 16th, 2017

Thank you for your honest approach but if you had not done the test, we would not know of the incompatibility with Premiere which is widely used.

Your test revealed some very useful information and even though the finger was pointed in error to the camera, I would hope that you do not stop testing new cameras.

It is tests like yours that carry more weight to companies like Adobe and Panasonic for useful change.

I just got the URSA mini pro and found out the 3:1 and 4:1 RAW is not supported by Premiere. When I asked Adobe about it, I was pointed to their wishlist site.

We need guys like you doing test like these to help us get these companies to give us the features we should expect to have.

 Guy McLoughlin
Guy McLoughlin
MemberApril 21st, 2017

Format incompatibility has been around forever, and always happens with the release of new cameras that can do new things. Wait a couple of months and every major NLE will be updated to open these new files.

Pawel Chyrowski
Pawel Chyrowski
GuestApril 16th, 2017

OMG, World doesn’t end on adobe, fcpx, davinci. Use Edius. Full support for 10 bit H264 without conversion.

Łukasz Królak
Łukasz Królak
GuestApril 16th, 2017

Adobe AE opens without any problems 10bit H264 ;)

Pawel Chyrowski
Pawel Chyrowski
GuestApril 16th, 2017

Open yes, but with the gamma shift on vlog :)

Łukasz Królak
Łukasz Królak
GuestApril 16th, 2017

I dont see any gamma shift on vlog in Adobe. Why this guy convert h264 to Prores? Nonsesne…

Pawel Chyrowski
Pawel Chyrowski
GuestApril 16th, 2017

How do you know, if you had only vlog? And yes, this is most funny thing, like convert mp3 to wave.

 Nicola Verdi
Nicola Verdi
MemberApril 16th, 2017

Do you know ONE post production facility using EDIUS? Did you ever been at Fotokem or Company 3? What a stupid comment…

 Edward Chew
Edward Chew
MemberApril 16th, 2017

Thank you

Richard Dimery
Richard Dimery
GuestApril 16th, 2017

Anyone understand why vlog clips the highlights at around 80% on the IRE. Seems either like a huge waste of data or deliberate dynamic range suppression.

Momir Alvirovic
Momir Alvirovic
GuestApril 16th, 2017

V-Log L is derivate of full Varicam V-Log, and it uses the same curve with cut off at 12 stops. So, V-Log L cuts off at aprox. 12 stops of dynamic range maxing at IRE 80 while full V-Log uses full scale for aprox. 16 stops of dynamic range… Please see these two links

https://suggestionofmotion.com/blog/panasonic-gh4-setup-v-log-overview/

http://pro-av.panasonic.net/en/dvx4k/pdf/ag-dvx200_tech_brief_vol6_en.pdf

Richard Dimery
Richard Dimery
GuestApril 16th, 2017

Much appreciated. It would be great if they made a version more optimised for the sensors and codecs of the lower end stuff. I appreciate why they have done this but I’m not going to need to match varicams for the most part. All it’s really doing is making the footage look worse.

Adam Loretz
Adam Loretz
GuestApril 16th, 2017

Interesting update. I look forward to everybody’s experiences and workflows using the GH5, it’s certainly not all figured out yet. I’ll be sharing my findings too ?

Dennis Ng
Dennis Ng
GuestApril 16th, 2017

Should continue.

Lukas Kalinowski
MemberApril 16th, 2017

Hi guys Thanks for test! Does any one compared 1080p 4:2:2: 10BIt from GH5 with Blackmagic Pocket Cinema Camera ? i found one short comparision on YT and probably due to bad settings on gh5 its looking ugly. I’m looking for update to my BMPCC. Does CINEMA 5D did dynamic range test in LAB on BMPCC ? i’m really interested how does this two cameras would compare in dynamic range and how Your test would measured this. Cheers Luke

 John Norton
John Norton
MemberApril 17th, 2017

There’s no FHD 10 bit 422 option available on the GH5.

 Guy McLoughlin
Guy McLoughlin
MemberApril 21st, 2017

Right now the GH5 can’t shoot this format, but Panasonic is releasing a firmware update on April 24th that will add 1080p 10-bit 4:2:2 recording.

 Ronald Evans
Ronald Evans
MemberApril 16th, 2017

Yes. Please do not stop doing reviews but maybe test files on both Mac and PC so that users can identify problems like the one you found with Premiere and ProRes. I use PC and have a GH5 , no Vlog but the 10 bit 4:2:2 files from the camera work in EDIUS WG 8.3 but not in Vegas 14 as an example. UHD files, 8 bit 4:2:0 work in both. Being able to use the files is in my opinion also a useful part of the testing.

 Stephan Pirsch
Stephan Pirsch
MemberApril 16th, 2017

Maybe, you could catch the opportunity and revise your NX1 review, too.

Jimmy Pitts
Jimmy Pitts
GuestApril 16th, 2017

how’s the autofocus in movie mode?

Johnnie Behiri
Johnnie Behiri
AdminApril 16th, 2017

Hi Jimmy.

For now I get mixed results. Will publish my full review towards the end of the week.

Thank you.

Johnnie

 Ronald Evans
Ronald Evans
MemberApril 17th, 2017

Test with Custom OFF and try the various ways of telling the AF what to focus on. I find that it behaves much like my Sony camcorders with Custom AF OFF and just Continuous ON. Has the advantage of specify the area or face detection.

 Guy McLoughlin
Guy McLoughlin
MemberApril 21st, 2017

GH5 AF is very dependent on your settings…
https://www.youtube.com/watch?v=OwDE92krDBo

 Konstantin Zettiness
Konstantin Zettiness
MemberApril 16th, 2017

If Premiere interprets your C700 prores footage correctly, then obviously it’s the conversion tool you were using — why not mention that to start with?

 Caleb Carges
Caleb Carges
MemberApril 17th, 2017

The issue is likely with how Premiere Pro (or Adobe software in general) interprets the colorspace metadata in the ProRes file. This may be why converted C700 footage works fine while GH5 footage doesn’t. Premiere Pro misinterprets BT.709 ProRes footage out of my Atomos Ninja Assassin recorder as BT.601 for instance.

 Cary Knoop
Cary Knoop
MemberApril 16th, 2017

Sebastian, could you provide the unaltered camera clips so everybody can check the problem?

 Nicola Verdi
Nicola Verdi
MemberApril 16th, 2017

Sebastian PLEASE don’t stop to do lab tests! People like you helping the community and point to problems. Its is so easy to write some negative critics but to spent hours in the lab and doing testing, looking for problems and provide some explanations is totally different story. By the way: Do you know why the English make better lovers than the Germans? Because English are the only one’s who can stay on top for 45 minutes and still come second ;)

Michaël Marcopoulos
Michaël Marcopoulos
GuestApril 16th, 2017

Not a real surprise

 the SUBVERSIVE
the SUBVERSIVE
MemberApril 16th, 2017

Thanks for the update but hopefully you guys will be less prepotent if you guys go back to lab tests. I say that because a lot of readers pointed out a possible codec problem and you guys talked like your methodology was flawless. A lot of people suggested a external recorder as a way to easily find out if codec was an issue but those suggestions in the comments were widely ignored.

I understand that an error like this was not expected, but then you guys can’t possibly state that this was not rushed in some ways, once you find an error like this you need to double check just in case.

So although I appreciate this update, I wished you guys would also admit that you guys should also be held accountable as part of what happened and not simply blame the software, mainly for the replies you guys threw at readers of this and other pages like 43rumors, where you guys also wrote comments. Not just that but there were a lot of comments by you guys bashing on people that actually questioned your methodology and also pointed flaws in it.

Don’t hold your pride above the matter, a bit of humility is appreciated.

 Kevin Almodovar
Kevin Almodovar
MemberApril 17th, 2017

He’s that guy.

 Guy McLoughlin
Guy McLoughlin
MemberApril 17th, 2017

Thanks for the updated review.

If you ever evaluate another DSLR/Mirrorless camera please include tests with an external recorder, as MANY of us shooters use external recorders to get the best possible image from our cameras, and it can make a LARGE difference in the quality of the final image.

You might even consider doing a specific article comparing several popular cameras when it comes to internal recording and external recording. Both the Sony and Panasonic cameras show a noticeably better image with external recorders.

 SL Huang
SL Huang
MemberApril 18th, 2017

I ran into the same problem when color correcting Luke Neumann’s test footage on my 2013 Mac Book Pro, in Premiere CC 2017. Transcoded to ProRes HQ and the graded result had noticeable banding. When grading the native codec it looked fine—but when exporting the corrected footage to ProRes, I ran into banding again. I definitely don’t blame Sebastian for the test results, as this is all very unexpected and weird. I hope it all gets ironed out with updates!

 Adam Muro
Adam Muro
MemberApril 19th, 2017

GH5 owner here. Firstly, Thanks for the tests! I would just add that, although the 180fps slow-motion is indeed somewhat underwhelming…the 120fps is actually quite lovely.

Benjamin Corwin
MemberApril 21st, 2017

Don’t sweat it Sebastian. There’s all kinds of wonky things about these smaller cameras. You have to be borderline obsessive to iron them all out. Keep up the good work!

 Petter Magnusson
Petter Magnusson
MemberMay 3rd, 2017

I have done some extensive test with GH5 vlog vs Cinelike D and 8bit vs 10 bit. Here are my findings:

vlog is way much more noisy than Cinelike D, I think vlog 3200 iso looks like cinelike D 800 iso when both are graded to look the same. However, vlog seems to be much less sensitive to Noise Reduction, I have Cinelike D NR at 0 and that is max for me to look good, while vlog can take NR +5 and still look sharper! And then the noise of vlog looks like the noise level for Cinelike D.

Conclusion for me: Anything above iso 3200 is unusable no matter what. For Cinelike D use NR=0, for vlog use NR+5

On 8bit vs 10 bit I have the strangest thing, if I do a test recording comparing 8 and 10 bit the 10 bit file has much more contrast and less dynamic range!

My settings are GH5, FHD, ISO 400, MP4 LPCM, Cinelike D with Sharpen -3, Saturation -2, Firmware 1.1, lum 0-1023/0-255, iDyn off, iRes ff, Highlight Shadow Standard

Premiere Pro CC (I have no problem playing 10 bit files FHD and 4K however!)

Can anyone reproduce this? Sure looks like a bug!

With vlog I also get more contrast using 10 bit, but there it will not limit the dynamic range so will only do good (less banding and artifacts when grading).

Conclusion: Use only 8 bit with Cinelike D and only 10 bit with vlog.

 Ari Kirschenbaum
Ari Kirschenbaum
MemberMarch 16th, 2018

I know this is an old post, so someone might have already pointed it out, but major mistake regarding the speed booster info you made in your original article.

A speed booster does not do anything to depth of field. It is placed after the iris therefore it is impossible to affect your depth of field. It increases FOV (correcting for crop)…and gathers more light, meaning you have the light of a faster lens but not the DOF.

1
1
- 3
Filter:
all
Sort by:
most voted

Take part in the CineD community experience

mode_comment40