Hi,
I have just tested the Colourspace LTE Rental version with the EIZO GG247X for 5 days. And I almost despaired because of one circumstance:

The LUTs showed visible banding with a grey gradient in Resolve also with 10 bit Timeline processing. A calibration with Colornavigator 7 shows the same gradient perfectly. At the end I found out that only the Displayport input can process 10 bit, the HDMI cannot.

I assume that the 3D LUT is only calculated cleanly with10 bits input. If I use CleanFeedOut in Resolve to output directly via the Displayport, the gradient is clean. My signal path is from Mac via an BM Ultrastudio 4K with SDI to a BM Miniconverter 6G and via HDMI to the EIZO CG247X.

- Am I seeing the problem correctly or does anyone have a different explanation or solution?
- Can I use the Displayport with 10 bit with an active HDMI to Displayport adapter or can this cause new colour problems?

Any help would be highly appreciated
Thanks

Daniel

The calibration process is to all intents and purposes bit-depth agnostic.
The LUT is always generated to the same resolution/granularity.
The the bit-depth of the patches really make little difference, as all that is important is the ColourSpace knows the patch values used for each colour, so it knows the expected value the probe should measure.
In that way, the bit-depth used for calibration is really irrelevant - so long as the there is no bit-depth conversion anywhere in the signal path.

As per the User Guides, Resolve is 8-bit only, and should only be used with an 8-bit timeline and signal path when used as a TPG.
That is a Resolve limitation, as BMD have not correctly performed the ColourSpace integration.
It may be better to use the iTPG within the monitor.
Or a true 10-bit TPG, such as PGen.

Do remember to test the LUT within ColourSpace too.
Both using the ability to apply any images to the LUT, as well as using Active LUT.

Steve

danielv

According to the EIZO Spec, the HDMI can take either 10-bit or 12-bit signals, so not sure you are thinking it is only capable of 8-bit.

Also, you could simplify your path by using the HDMI out on the Ultrastudio to the CG247X and remove the Miniconverter.

The CalibMate software will turn the Ultrastudio into a patch generator, you might find this a useful tool - https://www.lightillusion.com/calibmate.html

vlut, you are of course right. I omitted the Miniconverter and went straight from Ultrastudio to the monitor. And HDMI 1.3 can do 10 bit. But the problem still exists and seems to be in the colour processing of my EIZO CG247X.

I have tried everything and watched Stuart Pointon's video about 100 times. When I load the LUT into the monitor, it produces banding. If I load the LUT in Resolve under Video Monitor LUT and leave the CAL slot in the Eizo with the Unity Bypass only, I get a very good result.

Thanks Steve for your explanation, I have checked the whole signal path and have the same 8bit values everywhere, in Colourspace, in Resolve and at the output of the Ultrastudio.

Is there a known problem with the first generation EIZO CG247X colour processor or is my monitor possibly defective?
Or is it possible that a bug in the SDK is causing problems during the upload?

Thanks
Daniel

Please run verifications profiles of the LUT when active within Resolve, and when loaded into the EIZO.
Then a comparison can be made.
Ideally large 10^3 profiles, although a Grey Primary & Secondary Ramp+ would likely be ok.

Steve

Steve
Interestingly, the verification of the uploaded LUT shows good values, but I can see the banding on the screen. (See JPEG'S)I could only do a Grey Primary & Secondary Ramp+ as I only have an i1Pro2 and it takes a long time to do a Cube 10^3. I hope this works for you.

Im File Screenshots.zip findest du die entsprechenden JPEG's.
In the file Verification_and_LUT.zip you will find all verifications and the LUT.

Thanks a lot
Daniel

Attached file: Screenshots of the effect.

Attached file: All Verifications and LUT's

Sorry, German has sneaked in:
You will find the corresponding JPEGs in the Screenshots.zip file
Daniel

The 'EIZO_CG247X_SN21690076-i1Pro2-LUT_loaded_CAL10.bcs' profile shows major issues.
You have colours reading well outside the target gamut, as well as a very, very different RGB Balance.
Compared to the 'EIZO_CG274X_SN21690076-i1Pro2-UByp_CAL10-LUT_DR.bcs' profile the differences are obvious.

One of the reasons is the i1Pro 2 cannot read well in low-lights.

And when you have the LUT loaded in DR you have a much higher black level - 0.119 nits.
When the LUT is loaded in the EIZO you have a much lower black level - 0.027 nits

That means you have a signal path issue somewhere in your workflow...

So you have multiple issues.
Some signal path error of some type.
And the i1Pro 2 cannot read low-lights.

Those errors also mean your original profile used to generate the LUT is probably invalid too.

Steve

Thanks a lot Steve for your analysis. That sounds like a mess. I will have to repeat the whole test. Resolve is probably really not the right TPG.

Do you think the new Calibrite SL would give better results and is it supported by ColourSpace? I don't have the budget for the "right" equipment.

It's interesting that the i1Pro2 works well with Colornavigator 7. What I don't like about the calibration with CN7 is the blue shift in the shadows. That is also the reason why I am testing ColourSpace. With ColourSpace this aspect is better as far as I can judge from my faulty results.

Daniel

The levels issue is something that will likely also be happening when you are grading.
It needs to be understood and correct irrespective of calibration.
To start, check the black levels in the original profile - a LUT itself cannot alter black levels,

And yes, the i1D3 will be a lot better that the i1Pro 2 - use the i1Pro2 for probe matching.

And why CN7 appears to work with the i1Pro 2 is it is not reading many low light patches, and is not generating a 3D LUT.
It only reads a few patches, and generates a 1D LUT and 3x3 matrix.
And that is also why you have the blue issues in the shadows.

Steve

I need to look at the levels again. Actually, I don't have any problems in my daily work. I have limited the EIZO to video levels and in Resolve also the whole workflow in video levels. I may have set something wrong when profiling.

I will get a Calibrite Display SL. Am I right in understanding that the Calibrite Display SL is the same device as the i1D3 just with a new name? The change from xrite to Calibrite is causing confusion. I have also read here in the forum that you recommend the device up to 1000nits for a good lowlight measurement result. That would be the Calibrite Display SL.

As soon as I have the new measuring device I'll take another rental licence because a purchase is only worthwhile if I get a good result with my equipment.

Thanks for your support

Daniel

Any of the Calibrite i1D3 probes will work with ColourSpace.
That includes the SL.

What was the black level in the 'original' profile pre-LUT generation.
As I say, a LUT cannot change the black level, so having the LUT do that when loaded into Resolve vs. when in the EIZO is just wrong.

Steve

As soon as I am in the studio tomorrow, today is labour day, which in Switzerland is a day without work ;) I will check it.
Daniel

In the original profile, the black level is 0.0357 and the whole image actually looks better than with the loaded LUT. Did I do something wrong when creating the LUT? I used Fit Space and then adjusted it with PassBlack. I can't explain the higher black level with the LUT in DR at the moment either. However, the banding only occurs when I load the LUT into the Eizo. (see JPEG'S from my previous post)

I have attached the original pre LUT profile EIZOCG247X_3^3.bcs.zip.

Daniel

Attached file: Original pre LUT profile

The original profile is just a 3x3 cube, so there are only 3 patches in the grey scale.
Thar really is never going to work for an accurate calibration.
You would do better with one of the Grey Ramp profiles.

The black in the original is also approx. 0.01 nits higher than with the LUT loaded into the EIZO.
That is not possible, unless the monitor or probe is very unstable.
Most likely the probe, as it cannot read low-light values well, which is why it is not a good idea to use an i1Pro 2 for actual calibration.

Using Pass Black for Resolve will likely be the issue there...

Steve