| Forums | Register | Reply | Polls | Search | Statistics |
Display Calibration Light Illusion Forums / Display Calibration /  
 

LUT video scale PQ curve

 
Author anevard
ZRO
Male
#1 | Posted: 19 Aug 2021 21:38 
Is it normal for a LUT with a modified HDR colour space with 2390 tone mapping to curve back up after it is video scaled pass black
LUT before video scale pass black

LUT with videoscale pass black

Author Light Illusion


Male
#2 | Posted: 20 Aug 2021 07:28 
Yes, that is expected.

Steve

Author anevard
ZRO
Male
#3 | Posted: 20 Aug 2021 21:15 
Can I ask why?

Author Steve

INF
Male
#4 | Posted: 20 Aug 2021 22:04 
Because the 'active' part of the LUT is between 16-235.
Anything outside of that range is irrelevant, as it will not be processed/visible.
But the LUT will pass any such data, without clipping, enabling verification of any workflow issues.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author anevard
ZRO
Male
#5 | Posted: 22 Aug 2021 18:09 
Thank you Steve for the response.

The objective of the exercise is to determine how the Atomos Ninja works as a LUT holder with HDMI passthrough.

With SDR video scale was quite accurate when verifying a LUT.

HDR is a bit of a mess. I realize that there are a number of other issues, particularly the fact that I am using a JVC X790 projector, which I know is not as linear as one would like.

However, before I explore the other issues and the best way to optimize the PJ for home theatre use, I am trying to understand what is going on with the video scale issue and the Ninja.

PJ was initially profiled 17^3 with Colour Profile off, high bright with peak white adjusted close to 0.3127 0.3290 and 2.6 gamma.

Custom ST2084 rec2020 colour space was created with 5x 2390 static tone map for 67Nit peak and 1000 nits mastering.

LUTs were created using Map Space, Fit Space and Concatenation. All with similar results. I used the concatenated LUTs for this series of tests. Videoscale was applied to the final concatenated LUT.

Verification profiling was grey and primary ramp only for now. I verified the LUT in CS with no Video Scaling through a Unity LUT in the Ninja. I then verified LUTs in the Ninja with and without videoscaling. I am seeing, what I believe are, strange results with the Ninja.

I wanted to attach reports but CS is not responding when it tries to calculate gamut coverage. I let it go for 2 hours and no response.

I have attached 3 profiles:
Concat verif in CS NoScale
Concat verif Ninja VideoScale
Concat verif Ninja NoScale

Any help, ideas or thoughts would be greatly appreciated.

https://www.lightillusion.com/forums/shared_files/storage/Profiles/3mverificationbprofiles.zip

Author Steve

INF
Male
#6 | Posted: 23 Aug 2021 12:03 
It looks like the LUT is being distorted when loaded into the Atomos...
And it doesn't look like Video Scale is required.
(You have not provided the colour space you generated...)

And the gamut of the display is so bad, it is unlikely Gamut Coverage can be accurately calculated.
We'll look into adding a catch for such bad profiles.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author anevard
ZRO
Male
#7 | Posted: 24 Aug 2021 18:55 
Steve:
And the gamut of the display is so bad

Even though it is a JVC projector with 1700 hours on the bulb, something strange is going on. I am posting the initial 17^3 profile done less than 2 weeks ago. The gamut isn't wonderful but a lot better than the last verification profiles.

I am also posting my latest custom colour space profile. I have been playing with various P3 and 2020 ones.

I am going to take a fresh look at some of the Atomos settings.

https://www.lightillusion.com/forums/shared_files/storage/Profiles/hdri17x3.zip
https://www.lightillusion.com/forums/shared_files/storage/Profiles/rec2020v.0021-67e1000ngx.zip

Author Steve

INF
Male
#8 | Posted: 24 Aug 2021 20:19 
Not sure what I can say...
The profile is from the projector in its 'Native' colour space/gamut, and looks ok.
(Use Manage/Extract to then compare the profile to the projector's own native colour space.)
But is still an appalling gamut for HDR - for example green is way off hue for HDR, as well as low gamut...

Steve
Steve Shaw
Mob Boss at Light Illusion

Author anevard
ZRO
Male
#9 | Posted: 29 Aug 2021 19:53 
I seem to have found something interesting after going crazy and running every possible test I could think of. Although there still remain a number of issues with creating a good LUT for my limited gamut X790 projector, I seem to have found a solution to one of the problems.

All the 2390 Tone mapped LUTs that were video scaled pass black, had an issue at the top of the curve. Every single one. Those without video scaling did not have this problem, but after many tests, I can confirm that the Ninja V does need the LUT to be scaled.
2390 tone map and video scaled

Finally I decided to try the soft roll-off rather than the 2390 tone mapping. Et voila, it worked! There were no strange issues at the top of the video scaled LUT.
Soft roll off and video scaled

I am not sure what the issue is, but I at least have a solution for this problem and can now move forward to next steps.

Author Steve

INF
Male
#10 | Posted: 29 Aug 2021 20:19 | Edited by: Steve 
I think you are confusing what Video Scale does with the reprofiling of a display that doesn't require Video Scale, when a Video Scale LUT has been applied?

Steve
Steve Shaw
Mob Boss at Light Illusion

Author anevard
ZRO
Male
#11 | Posted: 29 Aug 2021 21:20 
Steve:
I think you are confusing what Video Scale does with the reprofiling of a display that doesn't require Video Scale, when a Video Scale LUT has been applied?

Both of the above profiles are with LUTs that are video-scaled and loaded into the Atomos Ninja V. The only difference is the target colour spaces that were created.

The top image is a LUT targeting a modified ST2084-2020 colour space that used BT2390 tone mapping, The bottom image is a LUT targeting a modified ST2084-2020 colour space that used soft roll off. The one with soft roll off follows the colour space target curve fairly closely. The one with the BT2390 tone mapping has that weird deviation at the top.

If I use non-video scaled LUTs and profile through them as active LUTs in CS, with a unity LUT in the Atomos, they both follow the target colour space curves pretty well. Only when the BT2390 LUT is video scaled and loaded into the Ninja does it profile with that weird deviation. The soft roll off LUT that is video scaled and loaded into the Atomos profiles similarly to its non video scaled counterpart loaded as an active LUT in CS.

Author Steve

INF
Male
#12 | Posted: 30 Aug 2021 07:29 
Sorry, As you have not supplied the profiles, LUTs, or the verifications, there is nothing I can say.
For example, how do the LUTs respond via 'Active LUT' vs. The LUT Box?
(With Video Scale applied, and if necessary Patch Scale set to match.)

But, everything still appears to point to an issue with the LUT Box...

Steve
Steve Shaw
Mob Boss at Light Illusion

Display Calibration Light Illusion Forums / Display Calibration /
 LUT video scale PQ curve

Your Reply Click this icon to move up to the quoted message


Watch this topic / Get updates by Email
 (?)
Only registered users are allowed to post here. Please, enter your username/password details when posting a message, or register first.

 

 
Online now: Guests - 1
Members - 0
Max. ever online: 312 [11 Jun 2021 17:33]
Guests - 312 / Members - 0