Greetings,
I am encountering a constant drift of blue when profiling HDR, see drift.jpg.
- Reproducable with three different probes (xRite i1Display (10y old), Calibrite Display Plus (new), i1Display (new)
- Same issue on a new PA32UCX-PK, and a used Cooler Master GP27U-FUS
- PGenerator on Raspberry Pi 4, connected via LAN cable
- 850 pre roll patches
- Stabilisation patch (black) for 2s after every patch
- Drift patch every 20 patches (75% grey)

SDR profiling does not show this issue, only when profiling the monitor in HDR mode

Anyone who can confirm this in a similar setup? Been troubleshooting for several days.

What delay value do you use, and which meter settings?

All FALD Monitors will suffer from such issues in HDR.

However, you can set a gray value as a background patch color to check if it stabilizes the panel's performance.

With FALDs you will get very different results using a small patch window, with any color bg, vs full screen patches.
I junked my FALD due to such issues.
But at least you can see and work with the Drift data with ColorSpace, so you know what you are up against.
And the LUT Generation will compensate fore the drift.
But the monitor will still suffer the same drift when in use.

Hi all,

Apologies, I just realized my post has been approved.

@ConnecTed Tried several combinations of values, none of them improved the issue. Mostly going with 1-3 seconds measurement, 0.5s additional delay. I always do 1-1.5s black stabilisation patch after every measurement. But also tested maximum values like 6 seconds for troubleshooting...

@ColorCal That is interesting. One, because I usually measure full screen patches with FALD displays. But good to know results may differ. Yes, it is good to have drift compensation in general. I am just totally surprised that the drift is so volatile... but I can accept it is a systemic issue with this technology and move on.

Also testing a PA32UCDM right now. Drift looks similar....

I cannot believe the PA32UCDM drift looks anything like the PA32UCX.
Are you sure you have not confused profiles?

Yes absolutely. However one thing was discovered: When profiling the UCDM in HDR, the brightness of the drift patches has a _massive_ impact, at least with this probe. It seems I had it too high (192,192,192) which is really bright in HDR, but may work fine in SDR.

With that the drift is significantly reduced. Together with a black 1s stabilisation patch I am now getting drift below 0.5

I have not yet tested with my GP27U FALD panel.

I still cannot believe the drift plot for the PA32UCDM looks anything like the plot you have posted for the PA32UCX.
One is FALD, the other a QD-OLED.
It is just not feasible the drift is similar.

Similar in sense of erratic and all over the place... please consider that this is not my profession and I am making my first steps with Colour Space ;)

See both example spaces attached as zip.

But that problem really appears to be solved. I am getting a lot less drift now with the darker drift patch color. Not sure what is happening there, but a bright drift patch color is producing an extreme drift, at least for me, at least for HDR.

Attached file: Archiv.zip

Drift patches should in most cases be around 90% of nominal peak white, although the display/probe combination can require different values, as well as different Pre-roll and Stabilisation patches.

For SDR, peak white is always 100% of the display's peak, so around 90 nits Drift patches for a calibrated grading display.
For PQ HDR nominal (diffuse) peak white is approx. 200 nits, so approx. 180 nits Drift.
For HLG the value will vary with the display's peak luma.

Steve

cru
Neither of those Drift plots are the same as the original PA32UCX Drift.
And they are also not the same as each other.
The GP27U shows major general color temp. drift over time, as well as huge random instability.
The PA32UCDM just shows random instability, with no obvious underlying general color temp. drift

ColorCal
Thanks & good to know. I cannot say for sure if in any of these cases it was the probe that introduced this, or the panel it was measuring. Though I have tested with three probes as mentioned. I believe the drift issue can be reproduced, simply by starting an HDR profiling with a 192,192,192 drift patch, and no stabilisation patches. I have sent back the PA32UCX-PK for several reasons (e.g. noise, temperature..) and will stick with the UCDM, so I cannot test that one anymore.

This kept me busy for some time, but is now solved, see attached.

Attached file: UCDM.zip