I'm at a loss here, and I wonder if anyone has encountered a similar problem, or has any advice to offer.
The jist of the problem is- the monitor clips details in the lower shadows.
When using the monitor with either the factory Rec.709 preset or after LightSpace calibration, several industry standard chip charts (including Resolve's SMPTE bars and LightIllusion's BrigtnessCal test image) show the lower shadows clipped (around 3 chips of the BrightnessCal, or the 3.5 IRE chip of the SMPTE), while the video scopes (luma waveform) clearly show the existence of data. The highlights don't clip at all (LightSpace's ContrastCal)
I have tried using both HDMI ports, both DisplayPort ports, changing the cables, changing the BM card (Intensity Pro 4K/Decklink 4K Extreme 12G), hooking the monitor directly to the PC (Windows 10- GTX 970). The problem has nothing to do with data ranges- no combination of Limited/Full ranges and YUV/RGB in the monitor menu or software yielded the missing shadow details.
I tried updating the monitor's firmware and ColorNavigator NX to the latest version (even switching to ColorNavigator), but the issue has not been solved. Other than that, the monitor performs admirably in the calibration and assessment using probe and software- both in terms of black level, contrast and gamut coverage (Rec.709).
Any ideas/similar experiences? I'm at my wits' end...

Hi Tomer,
You did say to me: "It seems to be a problem in the Blackmagic video cards' interaction with the Eizo, as I was able to get a proper range connecting it directly to the graphics card."

Is that still correct?

Steve

I re-tested this morning, and on my machine- same behaviour on the GTX 970 as the BMD Decklink. It may have jumped to another preset when I previously connected it to the computer, but I checked and double-checked - same results.

Hi Tomer,
Yes, I faced exactly the same issue with the same monitor model one of my customer has. We was not able to find the way to fix this..

Sounds like a display issue then - time to go back to Eizo and ask them for assistance...

Steve

Hi Vassily! Thanks for the info- was it a faulty specimen, or inherent to the entire model?
Steve- is that something that can be handled by a corrective LUT?
I have heard such praise for the Eizo's, I can't believe their top-of-the line is so flawed for video...

Also- same behaviour on the CG248-4K?

In theory using the Parametric Gamma capability of LightSpace could work...
But, in reality it would be very, very difficult to calculate the values needed.

I would get in touch with Eizo to see what they say.

Steve

Hi Steve,
Delving deeper into the matter, I got responses from users across the globe, and went to check out one of the few CG318's here in Israel. At this point, it seems all the CG318's behave the same way- it is not the case of a faulty sample. What bothers me is this- is there indeed a problem, or is the display behaving as it should in Rec. 709?
From my early days in video production, when I learned how to calibrate video monitors using color bars, the instruction was to use the PLUGE section in the SMPTE bars to set the brightness so that the 3.5 and 7.5 IRE patches disappear:
https://www.videouniversity.com/articles/color-bars-and-how-to-use-em/
https://www.youtube.com/watch?v=qkQjSAHxxi0 (from 2:17)
To quote: "Since 7.5 units is as dark as video gets, you should not see any difference between the left bar (3.5 units) and the middle bar (7.5 units). There should be no dividing line between these two bars. The only division you should see is between 11.5 and 7.5."
That is indeed what the display shows on the Rec. 709 preset/calibration (as opposed to the sRGB preset, where all the dark patches are visible).
Isn't that how a calibrated video monitor should display the dark tones?
I also saw a graph showing the luminance response in Rec. 709:
http://cdn.redsharknews.com/images/Rec._709_gamma_curve_shown_here_in_comparison_to_the_sRGB_curve_used_on_computers.png
(Taken from: http://www.redsharknews.com/technology/item/3289-what-exactly-is-rec-709)
Looking at it, it seems very obvious why the shadows should behave differently from the highlights.
So, my question is:
Is there indeed a problem in the display showing the bottom 3 patches in the LS BrightnessCal test image when calibrated to Rec. 709?

Pluge will work with TV Level displays, if the display passes below black signals.
The only way to prove this would be to test.

But, the Pluge signal is a lot more 'course' than the BrightnessCal image... so although the Pluge may look correct, the BrightnessCal image may easily still show clipping. It is why we use the BrightnessCal image for our calibration - to show such issues that Pluge will not show.

I would talk to Eizo, and ask them why this is happening?

And the Redshark graph is basically bollox!
Rec709 uses a standard power law gamma, and looks nothing like that graph, which is very confusing to say the least!
They do not even mention the gamma used, as Rec709 has no pre-set gamma value, and 2.4 would be very different to 2.2...
And at no point would the signal be clipped, as the graph suggests - that really is just wrong!
(There are some serious issues with that whole piece, so I'd suggest not believing much that it says - unfortunately we have had issues with Redshark before, as they seem to publish info without actually checking it for accuracy - a search on LGG will show this.

Steve

A reach-out to Eizo would be greatly appreciated, as they don't seem to communicate much with mere mortals
I see what you're saying re:redhsark, they do sometime seem more prosumer then professional.
I do not know what is the gradation in BrightnessCal and how it compares with IRE levels, but it definitely not a data_vs_video error, as there is a lot of visible data under 16 (only 3 patches clipping). It may be clipping under the 7.5 IRE- again, I can't reach that conclusion on my own, and more importantly- I don't know if that is acceptable, or problematic.

The patches in the BrightnessCal image are 1 bit level changes (in 8 bit) along the top row, and 2x changes down the left hand side.
So, 1,2,3,4,5,6,7,8 along the top
And 1,2,4,8 down the left hand side.
(The second row is 2 bit level changes, so 2,4,6,8,10,12,14,16... and row three is 4 bit level changes... and row four is 8 bit level changes).
(You can use http://www.iconico.com/colorpic/ to sniff the values on any image!)

I have sent a message to Eizo, so hopefully they will respond.

Steve

Hello, this is Victor from EIZO UK, before I jump in and try to answer this question will check wth our R&D team in Japan as this is not a concern that has been raised.

Would I be correct in thinking you are setting the luminance/brightness to 100 cd/m^2 and the gamma to 2.4 or using a LUT?

Cheers

Victor

Hi Tomer.
This conversation thread came to me via Penny at Eizo Oceania, based here in Sydney, Australia.

I spent two days locked in a dark room with the CG318-4K and the smaller CG248-4K late last year, and banged my head against the table many times with this same issue.

When using Eizo's own ColorNavigator NX software to calibrate these monitors, the absolute black level of the monitor is 'lifted' slightly when the software is actively communicating to the monitor. Try this - If you connect your PC to the monitor via USB, feed it Steve's BrightnessCal test chip image from the same PC or from your SDI/HDMI video card of choice, then launch the CN NX software. You will see the black level lift when the s/w is talking to the monitor, and it will drop again when you exit CN NX.

Note: You don't necessarily need to see the CN NX software UI on the Eizo screen to do this simple test, but you will do if you intend to run a calibration using the CN NX software.

This lift up/drop down of blacks may or may not be directly linked to your shadows clipping, as I don't know for sure how you were setting the base calibration in your case. CN NX or LightSpace ?. But in my own work using CN NX for my (non-profiling) calibration work I suffered the exact same problem on both the 30inch and 24inch models. The solution is very simple though....

Create yourself a bunch of alternative LUTs in LightSpace that lift the blacks. This is done in the Edit > LUT Manipulation > Lift menu.
After making your calibration, load whichever of these Lift LUTs into the monitor via the CN NX software such that it brings up the shadow detail to the point where you are just seeing the darkest black chips (as you would on any other broadcast monitor). I think the button in the CN NX to load the LUT is called 'Log Emulation'.

If you are loading a fully profiled 3D LUT created by LightSpace CMS, then you may need to combine the Profile LUT with your Lift LUT, and load the two as one single LUT.

I found that for a 100Nit peak white I needed to apply around 3% lift, and for a 48Nit peak white I needed to apply a 5% lift.

If this was a traditional 'video' monitor it would have a knob or menu called Brightness to do this. But it's not, and so you have to do it via a LUT. Luckily the Eizo has a place for this LUT, most don't, and it will even allow you to load a different LUT for each and every Mode Preset.

I'll also share another finding from my work here. The CG318-4K and CG248-4K monitors are different and give different results. The backlight system is different, and so don't assume that any custom work you do on the 318 can be directly applied to the 248 and will still achieve the same results. The technique can be applied to both, but you most certainly need to treat each monitor individually.

Let me know how you go.

Stuart Monksfield - Mojo Media Solutions, Sydney, Australia

Hello Tomer,

Can I ask you to try the settings in the file linked here:
https://www.dropbox.com/s/989cpvfrvcvo8fy/REC709%20RBL%3DTrue%20BT1886%20100cd%206500K%202.40.cntarget?dl=0
It is an XML calibration target for ColorNavigator 6 and you can import it from the advanced button. I have done a test with this and I can see down to gray level of 5, below which is invisible for me.

Let me know how you get on?

Cheers

Victor