Questions about drift

Do the following

• Create a new profile with DCI P3 D65 as Target Gamut, 8bit limited rainge
• Select Virtual Probe with OETF DCI P3 D65
• In the profiling tab, enable Drift measurements, RGB level 64,64,64, Frames 1 (one drift measurement after eacu patch mesurement
• select fullrange (provided) as custom profile, with sorting sequentiell, and press "Start" to measure it
• change virtual probe to OETF DCI P3, select 20StepGrayIntermediate as custom profile, select "Add/Modify Points" and press "Start" to measure it
• look at the RGB and Diff EOTF graph, they look very resonable (screenshot Drift1)
• then select "Drift" in the graph options
• RGB and DIFF EOT graph look very strange now (screenshot Drift2), is this correct?
• Change virtual probe to OETF DCI P3 D65 back and measure again
• look at the RGB and Diff EOTF graph. One probably would expect flat lines now, since measuremenst are identical to the reference now, but this is not the case. (screenshot drift3)
• then deselect "Drift" in the graph options
• and the graphs look again as expected (screenshot Drift4)

I wouldn't say that this doesn't happen in practice; I've observed this strange behavior in the following situation: I wanted to record an HDR profile, and to reduce thermal drift, I mounted the measuring device on a tripod, placed it about 3 mm in front of the display, and set up a USB fan next to it for cooling. Unfortunately, the fan stopped working shortly before the end of the measurement, so I remeasured the last 200 patches.

Attached file: The first patch sequence	Attached file: The second patch sequence	Screenhot 1
Screenhot 2	Screenhot 3	Screenhot 4

You are trying to apply a Drift sequence that is out of step with the data.
Drift is a weighted average over time.
Basically, it will never work as you are expecting.
You just cannot us it like that.

Steve

Does this mean one should never remeasure individual patches when measuring drift?

As per the User Guides, Drift is really aimed at being an assessment of the issues faced by the display's Drift during profiling.
Using the Drift values for aby actual processing of profile measurements is not really recommended.
It should be considered a last resort if the displays has really bad Drift that cannot be tamed.

So no, trying to use Drift and re-measuring patches is not a good idea.

https://lightillusion.com/hint_drift_aug.html

Steve