Unfortunately I am not sure what exactly may changed this because a technician swapped my main-board not long before the last EU firmware updates (which also changed the TV's behavior a lot...). And I don't really had either too much free time or even will to check this even further than doing simple re-calibrations after these changes (main-board swapping, firmware updates, etc...). Nor I checked it this deeply for the people who I made calibrations for recently. (Why would I always individually check this other than the obvious sanity checks if the charts look good and I evaluated the display deeply once...? - But I will do it next time...).
But I checked it today and it turned out I have terrible color accuracy.
First I checked the CMS quality and I found this:
This shows the CIE x,y coordinates of the primary colors (Native and Custom gamut + Rec709 reference). It is obvious that my Custom gamut settings isn't really applied constantly for the whole color gradients. They are only valid for the point where I calibrated the gamut (92.2%). And I always knew the CMS isn't perfect but this is too much...
I profiled the display with a simple gamma+matrix profile with IRE92 color measurements (where I calibrated) and validated the profile with random mixed colors. I ended up with huge dE2000 errors (dE~6 peak error
). Not a surprise after the above graph but still worse than I though...
I freaked out, switched to PC mode, reseted every settings (including the RGB Gains in the Service Menu) and calibrated the display as a PC monitor. The VGA LUT calibration resulted in a nice gray-scale, but the validation of a simple matrix+gamma profile failed with dE2000=5.8 peak and average ~4.8
This is not good!
I remember I experimented with this kind of calibration (as a PC monitor in PC mode) before I got informed about the D550@D7000 hack and moved to the internal CMS (10p gray and 6p gamut). I don't have the measurement data anymore but I also evaluated the internal CMS this way (calibration with the hardware controls and gamma+matrix profile validation on random mixed colors) and I remember that dE2000 values were very good (may be peak error ~1.7 close to cyan but average about 1.0 or so...).
If they were this high at the first time (PC mode calibrated as a PC monitor) I would send it back to the shop until I could (in the first ~13 days or so...). But I was happy about the great color accuracy. And now, I can't get accurate colors with this TV, no matter what.
Any ideas...? Could somebody else do the same measurements (10+ point on primary color gradients - I can explain how if you want but you don't know how...).
I will try to restore everything back to factory (including the model number) and repeat my initial tests (like when I unboxed the TV at the first time) as close as I can (from memory).
But I don't know what to do if I can't get similarly low dE2000 values again.
Well, Ok, I watched some movies and I didn't blamed about the colors. But that's because as I said, I didn't have too much free time, so I didn't want to spend it all with checking things like this and even though I thought I may have problems, I decided to ignore them (at least until today when I was in this mood to make some measurements).