Batutta
Major Contributor
Try Art Adams LUT for Sony F cameras. I stumbled upon it by accident. It's nearly a perfect match for VLOG and really reduces the artifacts--
http://www.provideocoalition.com/images/Art Adams AA709A LUTs.zip
GH4 Exposing V-LOG L
- Thread starter Bern Caughey
- Start date
Try Art Adams LUT for Sony F cameras. I stumbled upon it by accident. It's nearly a perfect match for VLOG and really reduces the artifacts--
http://www.provideocoalition.com/images/Art Adams AA709A LUTs.zip
ozmorphasis
Well-known member
Thanks or that. Just tried it. No artifacts in my quick test, but there was a slight cooling (and perhaps a bit greenish tint) to the white walls in my room that is not present to the naked eye. Needed to adjust a bit to balance, but otherwise very pleasing rendering, and most important, skin looks good!
In any case, definitely no magenta issues.
Batutta
Major Contributor
It seems to be making it a lot worse. Also, the noise doesn't seem as bothersome either. Start with the AA709A_Native_General LUT. I'm hoping an LUT that is exactly calibrated for the GH4 is released, but this one is nearly perfect.
Could it be a bit-depth issue ? ( What is the bit-depth that the Varicam 35 records in ? )
Erik Naso
Veteran
The Varicam LUT is for V Log from the Varicam so yeah it could be causing more problems than helping the grading processes. It's clearly not working with everybody having the same issues. Panasonic should release a LUT for the GH4 and V LOG L. I think that would at least get us in the right place. I want to play with the AA709A_Native_General LUT against the Varicam LUT to see how it measures up on the footage from today but I have a cold cocktail and some food I need to attend to first. :grin:
Batutta
Major Contributor
I'm starting to wonder, now that there is a better LUT, wether or not NR at 0 might be contributing to the magenta/green blotch issue, which is still there faintly in some of my clips if you push the grade. Will do some tests at -5 for NR and Sharpness to see if it goes away completely when I have the chance.
Erik Naso
Veteran
If this magenta issue can be fixed I would be pretty happy with V-LOG L. Color is nice. Skin looks good. Lets smash the magenta!!!!
Brandon Freeman
Well-known member
So I've been playing with FilmConvert, using the CineLike D Low Contrast profile in conjunction with KD 5213 Vis3. I know the color science is wrong but I seem (so I did a bit of color correction in the plug-in) to get very little magenta blocking (note the wall by the flower) compared to V709. Film grain also helps... not ideal for a lot of situations but for mine? I'm digging the look, I like FilmConvert's grain. 
https://www.dropbox.com/s/pwtq7qhisxgx/Flowers.png
EDIT: Am I not supposed to post Dropbox links? My link URL has an admin note attached to it - "I AM SPAM PLEASE REPORT ME"... Huh?
https://www.dropbox.com/s/pwtq7qhisxgx/Flowers.png
EDIT: Am I not supposed to post Dropbox links? My link URL has an admin note attached to it - "I AM SPAM PLEASE REPORT ME"... Huh?
Last edited:
I'd guess that hoping for a LUT that remedies this is a vain hope...especially in 8 bit
Unfortunately its in both LUTS...I just did some tests, here is a close up of cream walls in mid light that clearly shows the magenta blotches.
Its in the original untreated file too..and thats the problem, all the LUTS do is bring it out, you can see it (although harder) in the original file by pushing contrast and color.
I have attached closeups of the Varicam and the AA709A_Native_General LUT as well.
Left is the AA LUT right is the Varicam
Unfortunately its in both LUTS...I just did some tests, here is a close up of cream walls in mid light that clearly shows the magenta blotches.
Its in the original untreated file too..and thats the problem, all the LUTS do is bring it out, you can see it (although harder) in the original file by pushing contrast and color.
I have attached closeups of the Varicam and the AA709A_Native_General LUT as well.
Left is the AA LUT right is the Varicam
Last edited:
visceralpsyche
Well-known member
I'd like to see the root cause of the problem addressed by Panasonic, but I fear that's not going to happen.
From my testing, the root problem appears to be the first stage sensor processing, before it's sent to either external 10 bit 4:2:2 or internal 8 bit 4:2:0.
Since we know the camera's sensor is capable of around 12 stops from RAW photo tests, the problem lies between the RAW image capture and the video output chain.
It seems to me that the video encoder is not processing the image with a high enough bit depth before sending it for internal or external processing. I say this because in my 10 bit 4:2:2 testing yesterday, I still noticed the macroblocking. Given that I was supposedly using the uncompressed, 10 bit 4:2:2 output, and recording in ProRes, I shouldn't be seeing such processing artifacts.
My conclusion based on that evidence would suggest that the camera's 10 bit 4:2:2 output through HDMI is not in fact unprocessed at all, but rather processed down before output in a way that is manageable by the camera's CPUs.
This would also explain the external monitoring lag present in the GH4.
I'm actually really glad that V-Log was inadvertently released through the Image App, as it's helped me to understand that the camera is a lot more limited in its external output than Panasonic has been saying it is.
It could also help to explain how other 8 bit 4:2:2 external output cameras are giving a better image than the GH4's 10 bit output, if the other cameras are sending a true uncompressed signal using the full available bit depth over Panasonic's hobbled output.
It also makes me wonder if the GH5 will correct this, or whether the fundamental design is broken for the image chain.
Would love to hear other people's thoughts on my theory!
Cheers
From my testing, the root problem appears to be the first stage sensor processing, before it's sent to either external 10 bit 4:2:2 or internal 8 bit 4:2:0.
Since we know the camera's sensor is capable of around 12 stops from RAW photo tests, the problem lies between the RAW image capture and the video output chain.
It seems to me that the video encoder is not processing the image with a high enough bit depth before sending it for internal or external processing. I say this because in my 10 bit 4:2:2 testing yesterday, I still noticed the macroblocking. Given that I was supposedly using the uncompressed, 10 bit 4:2:2 output, and recording in ProRes, I shouldn't be seeing such processing artifacts.
My conclusion based on that evidence would suggest that the camera's 10 bit 4:2:2 output through HDMI is not in fact unprocessed at all, but rather processed down before output in a way that is manageable by the camera's CPUs.
This would also explain the external monitoring lag present in the GH4.
I'm actually really glad that V-Log was inadvertently released through the Image App, as it's helped me to understand that the camera is a lot more limited in its external output than Panasonic has been saying it is.
It could also help to explain how other 8 bit 4:2:2 external output cameras are giving a better image than the GH4's 10 bit output, if the other cameras are sending a true uncompressed signal using the full available bit depth over Panasonic's hobbled output.
It also makes me wonder if the GH5 will correct this, or whether the fundamental design is broken for the image chain.
Would love to hear other people's thoughts on my theory!
Cheers
TheTrickster
Well-known member
That is bad news if true (not doubting you)
Erik Naso
Veteran
I'd like to see the root cause of the problem addressed by Panasonic, but I fear that's not going to happen.
From my testing, the root problem appears to be the first stage sensor processing, before it's sent to either external 10 bit 4:2:2 or internal 8 bit 4:2:0.
Since we know the camera's sensor is capable of around 12 stops from RAW photo tests, the problem lies between the RAW image capture and the video output chain.
It seems to me that the video encoder is not processing the image with a high enough bit depth before sending it for internal or external processing. I say this because in my 10 bit 4:2:2 testing yesterday, I still noticed the macroblocking. Given that I was supposedly using the uncompressed, 10 bit 4:2:2 output, and recording in ProRes, I shouldn't be seeing such processing artifacts.
My conclusion based on that evidence would suggest that the camera's 10 bit 4:2:2 output through HDMI is not in fact unprocessed at all, but rather processed down before output in a way that is manageable by the camera's CPUs.
This would also explain the external monitoring lag present in the GH4.
I'm actually really glad that V-Log was inadvertently released through the Image App, as it's helped me to understand that the camera is a lot more limited in its external output than Panasonic has been saying it is.
It could also help to explain how other 8 bit 4:2:2 external output cameras are giving a better image than the GH4's 10 bit output, if the other cameras are sending a true uncompressed signal using the full available bit depth over Panasonic's hobbled output.
It also makes me wonder if the GH5 will correct this, or whether the fundamental design is broken for the image chain.
Would love to hear other people's thoughts on my theory!
Cheers
This all does make sense. Especially the lag issue. However it lags in all the camera settings but if it is because the cameras processor is struggling then that could be an issue that cant be fixed.
As for the 10bit out. How can we actual find out if the signal is uncompressed 10bit 422? I've recorded 10bit with Natural profile and haven't seen this magenta blocking I think V-LOG L just isn't a well tuned LOG profile and needs to go back to the engineering bench. It will be interesting to see after my key code arrives and I update the firmware to use it if the magenta issue still exists. I doubt it will go away. Why hasn't the beta testers said anything about this? Did they not see it? Is it an issue with the way the profile is implemented on firmware 2.3? Nick Driftwood shot a bunch of footage and released frame grabs we all played with and never saw this issue nor did we when playing with it.
Erik Naso
Veteran
Well that sucks. V-LOG L needs to go back to beta for some tweaking. Only thing left now is if v 2.4 firmware and V-LOG L actually fix this issue. I doubt it. With Panasonic saying the Image App release of V LOG L isn't an official release is also strange since the profile lived in the firmware. I'll have to wait for my key code but if anyone gets it soon please do a before and after to confirm if the issue is caused by the Image App backdoor release.
As someone who doesn't own this camera and reads this forum posts for the general information, I'd like to express my gratitude to those who have tested the camera fairly and disclosed their findings promptly. It's the folks like you who prevent these giant companies from getting away with this sort of crap.
Kudos!
PS. Not all companies are giant. When AJA Cion had similar magenta problems, the DVX User posters were not shy about what they saw either.
Kudos!
PS. Not all companies are giant. When AJA Cion had similar magenta problems, the DVX User posters were not shy about what they saw either.
Last edited:
visceralpsyche
Well-known member
Here's my test setup from yesterday.
Note the PIX-E5 input settings - 10 bit 4:2:2 @ 4096x2160x24.00p
The camera was most definitely set to output 10 bit 4:2:2 through the HDMI. Here's the wall clip I posted about earlier, to show that even a high key defocused image causes macroblocking on the 10 bit 4:2:2 output of the GH4:
http://downloads.visceralpsyche.com/pix_e5_vlog_test/PIX-E-001-009.mov (486MB download, 10 bit ProRes 422 HQ)
You can then see the macroblocking and play with your editing software like I did to try to mitigate/eliminate the effect as much as possible. It can be hidden, but it doesn't change the fact that the original 10 bit output is putting it there in the first place, unfortunately.
The clamping of the image signal (another reason for macroblocking due to limited bit bandwidth) can be seen in this photo taken by one of the Video Devices reps:
Note the yellow false colour on the PIX-E5 screen. That was me trying to get it to go to red (which signals over 100%). It wasn't possible thanks to clipping at 80% output signal, so we are essentially losing two stops of high end colour information and one stop at the low end. Hardly the best use of the 0-100 IRE output the camera is capable of in other linear modes.
I do wonder - did any of the beta testers see these kinds of issues? If so, did they tell Panasonic and get ignored? Or did they miss the issues entirely (even more worrying if you are a beta tester)? Because I would have been complaining long and hard about the waste of bandwidth at the top and bottom end, and the fact that no external recorder was recording the full bandwidth the camera output is supposed to be capable of.
Cheers,
Paul
Note the PIX-E5 input settings - 10 bit 4:2:2 @ 4096x2160x24.00p
The camera was most definitely set to output 10 bit 4:2:2 through the HDMI. Here's the wall clip I posted about earlier, to show that even a high key defocused image causes macroblocking on the 10 bit 4:2:2 output of the GH4:
http://downloads.visceralpsyche.com/pix_e5_vlog_test/PIX-E-001-009.mov (486MB download, 10 bit ProRes 422 HQ)
You can then see the macroblocking and play with your editing software like I did to try to mitigate/eliminate the effect as much as possible. It can be hidden, but it doesn't change the fact that the original 10 bit output is putting it there in the first place, unfortunately.
The clamping of the image signal (another reason for macroblocking due to limited bit bandwidth) can be seen in this photo taken by one of the Video Devices reps:
Note the yellow false colour on the PIX-E5 screen. That was me trying to get it to go to red (which signals over 100%). It wasn't possible thanks to clipping at 80% output signal, so we are essentially losing two stops of high end colour information and one stop at the low end. Hardly the best use of the 0-100 IRE output the camera is capable of in other linear modes.
I do wonder - did any of the beta testers see these kinds of issues? If so, did they tell Panasonic and get ignored? Or did they miss the issues entirely (even more worrying if you are a beta tester)? Because I would have been complaining long and hard about the waste of bandwidth at the top and bottom end, and the fact that no external recorder was recording the full bandwidth the camera output is supposed to be capable of.
Cheers,
Paul
Last edited:
Brandon Freeman
Well-known member
I'm not giving up on this yet. Something has got to be up with either the firmware or the workaround is causing some kind of issue - I just flipped through the 10-bit V-LOG L preview footage posted months ago and there's none of this magenta blocking anywhere. Either other testers are somehow making mistakes with their external recorders, or the beta testers flat out didn't have this issue.
Which means Panasonic can fix this.
Which means Panasonic can fix this.
visceralpsyche
Well-known member
I just checked the Shogun clip released from back then, and the magenta blocking is present and accounted for:
Since we know this was Shogun footage, and since my own PIX-E5 testing shows the same macroblocking in 10 bit 4:2:2, I have to conclude that the artifacts are embedded in the signal no matter which option you choose to record it with.
Both recorders appear to be faithfully recording the image presented to them through the 10 bit 4:2:2 HDMI output.
Which means my earlier post holds regarding my theory that the image is being degraded before we have a chance to record it.
Cheers,
Paul
Last edited: