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

Bad RGB separation when calibrating from Resolve

 
Author iali
User - Mono 
#1 | Posted: 9 Mar 2020 22:09 
Hello,
I'm having trouble figuring out my display behaviour when using resolve and a ultra studio mini monitor module as my patch generator. I have resolve output setup to Video levels and 8 bits per the recommendations in the guide. When profiling the display pre calibration with the i1 display OEM, I see very bad RBG separation in the quick profile primaries/secondaries run. When I use the supplied calibration patch java client on the same display but going through the OS display (Linux) with no ICC profile loaded(and what I believe is a null VCGT), I get much better results for RBG separation and overall lower Delta E.

Any thoughts as to why this is happening?

Author Steve

Mod - Scene 
#2 | Posted: 10 Mar 2020 06:37 
You'll need to provide the profiles so we can assess the results.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#3 | Posted: 10 Mar 2020 07:08 | Edited by: iali 
Thanks Steve. I've attached an archive with the profiles in question.
Just to be clear here's how they were generated:
*34um95 10cube video 8 bit:
LightSpace CMS VM->Resolve Patches on the same laptop->Ultra Studio Mini Monitor->LG 34um95 Monitor via HDMI

*PreCalib PRIMSEC JAR
LightSpace CMS VM-> Java Calibration Client running in full screen on Linux workstation with null profiles->LG 34um95 Monitor via HDMI to the workstation

Another related question: converting from Rec709 to the 3^10 Cube LUT or any variations generated through Resolve and direct output appear to produce much warmer shadows than the ones generated by the the Java client/OS. This is consistent regardless if I'm viewing the content through Resolve's dedicated output or through the GUI screen(from LightSpace Laptop or workstation).

When looking at the Calbrightness patches for example, the visible patches appear much warmer(almost reddish) as compared to the uncalibrated monitor and the LUT generated through the Java client. I know Rec709 probably has a warmer tones than the default on my crappy monitor so I'm not sure if I'm seeing the right thing here.

Profiles.zip Attached file:
Profiles.zip

 

Author Steve

Mod - Scene 
#4 | Posted: 10 Mar 2020 09:45 
Both profiles are hugely different.
Basically, there is something distorting your video path when you are using Resolve.
But, no idea what. That is something only you can define, as it cannot be remotely diagnosed.

And because of that, no calibration (LUT Generation) will work...
The profile data is basically invalid.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#5 | Posted: 10 Mar 2020 13:07 
I suspect there is something perhaps wrong with the Ultrastudio monitor because there is nothing else between resolve and the monitor, nothing much to setup on it though so I don't know what to troubleshoot.

Would you say the second profile from the Java client looks more realistic?

Author Steve

Mod - Scene 
#6 | Posted: 10 Mar 2020 13:14 
It could be a 'colour space' setting on the Resolve timeline?
Or as you say, in the output card...

But yes, the direct profile looks as I would expect

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#7 | Posted: 10 Mar 2020 14:23 
Thanks Steve, working space is resolve(so linear basically) and output is Rec709 2.4. I'm going to try and downgrade resolve to see if there are issues with the current version. I'm basically following your guide in setting Resolve so this is very strange.

Do you know of any other way of producing patches from LightSpace directly to something like the ultrastudio in order to make sure I'm bypassing all OS related color shifts?

Thanks for looking into my profiles. I really appreciate all the support!!

Author Steve

Mod - Scene 
#8 | Posted: 10 Mar 2020 15:06 
I would suspect there is some colour management set within Resolve. There are no known issues with any Resolve version, other than incorrect settings.

When using a dedicated video output card (not a graphics card) there are no ICC issues, as they have no effect - so no OS colour related issues.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#9 | Posted: 10 Mar 2020 15:23 
Thanks again Steve!!

I've attached screen shots of my current settings from the LightSpace laptop. I've followed the guidance from the "Resolve TPG issues" page, making sure output is set to 8 bits. On the LightSpace side I'm using 0-255 levels for the patches which resolve converts to video levels per the monitoring settings. I've verified what my monitor expects through the relevant cal images and can confirm that the images get clipped badly if I set resolve to Data levels so I assume my output settings in Resolve are correct.

There is no specific instruction on setting colour spaces(apologies if there is and I totally missed it), but If I'm starting to understand how things are supposed to work, could it be that I have to have my output/timeline colour space set to Linear so Resolve doesn't attempt to remap the gamut to Rec709 prior to producing the patches? This is the only thing I can think of.

Resolve Settings.zip Attached file:
Resolve Settings.zip

 

Author Steve

Mod - Scene 
#10 | Posted: 10 Mar 2020 15:54 
Sorry to say we are not Resolve experts..
You really need to talk to some Resolve users.
But Rec709 is not linear - it's a gamma power law.
However, the issue you have is not Gamma - it is some kind of colour distortion.

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#11 | Posted: 10 Mar 2020 16:43 
I went ahead and bought LightSpace Connect to see what's going on and I believe what I'm seeing with the attached quick profile makes more sense than what was generated by Resolve. Am I correct?

On resolve I noticed that beyond Green channel values of 230 nothing happens on the display. I guess Resolve is really messing with the colors somehow.

Thanks again for all the awesome pointers and support!

LS Connect TPG.bcs.zip Attached file:
LS Connect TPG.bcs.zip

 

Author Steve

Mod - Scene 
#12 | Posted: 11 Mar 2020 00:24 
Can't check until.later today, but does it look the same as the JaveApp profile?

Steve
Steve Shaw
Mob Boss at Light Illusion

Author iali
User - Mono 
#13 | Posted: 11 Mar 2020 00:53 
Rgb separation is off the diagonal but at least the 3 color curves are much tighter with no green clipping. But still not as good as the Java client. I'm assuming LS Connect doesn't suffer from ICC like issues?

Author iali
User - Mono 
#14 | Posted: 11 Mar 2020 17:18 
Quick update on my issue: It seem that the UltraStudio Mini Monitor has some issues identifying display capabilities for some reason. I was desperately trying different settings on Resolve thinking it was a colourspace mapping issue when I decided to switch over to my projector to asses the clipping in the green channel at least visually. As expected there was no clipping in the values when I did a manual measurement and moved the green channel sliders in the 230-255 range both visually and from the probe. When I switched the cable back to my monitor I happened to have the BarsAndBlack picture up on the resolve timeline and I could literally see the green bars on the top "pop" and I could actually differentiate the colours within the green box whereas I couldn't before. Obviously running a quick and full profile after this happened resulted in values that are extremely close if not identical to the Jar client and to a certain extent LightSpace Connect. Producing a rec709 calibration LUT based on these results resolved the extreme warm tones I was seeing in shadows with the bad profiles.
I read somewhere on the resolve forums that this output module has some EDID issues, I didn't think EDID mattered with this kind of output device because the display output settings are determined by the settings from the driver of that module. Does this make any sense to anyone?

In all cases, moral of the story: Use the BarsAndBlack calibration image to check that the display is behaving somewhat correctly and troubleshoot the signal chain before attempting anything else! Nothing new here that's not mentioned in the guides but I guess I had to learn the hard way

Steve, once again thank you so much for your continued support throughout my troubleshooting attempts!

Display Calibration Light Illusion Forums / Display Calibration /
 Bad RGB separation when calibrating from Resolve

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

 

 (?)
Only registered users are allowed to post here. Please, enter your username/password details upon posting a message, or register first.

 

 
Online now: Guests - 2
Members - 0
Max. ever online: 86 [28 Mar 2020 01:31]
Guests - 86 / Members - 0