Originally Posted by royboy365
Mr. Bob from Image Perfection spent some time at my place calibrating both my 9800 & 9300 and in the process we discovered a very interesting issue.
We were working with getting the proper Gray Scale and using his ISF Optical Comparator along with an Accupel HDG 3000 HD Pattern Generator. The pattern we were using had 11 bars from 0% to 100% inclusive.
On the 9800 whenever we adjusted a bar to the proper shade and exited the menu system, whatever change we made was showing on the next IRE percent up!
Easiest way to check this for yourself is adjust red, blue or green plus or minus 50 to make an easily observable change and then exit the menu. You will see the adjustment you made jump to the next highest level.
It did this on both the 9800 and the 9300. It also did it on the 9300 both on the previous software and the latest updated software.
Here's the full story. I just started a new thread on it.
Had a chance to work with 2 brand new LG OLEDs, and they both presented exactly the same anomaly: their settings didn't stay where they were put!
We used the TVS Pro Optical Comparator, with the ISF optical comparator riding shotgun, and my Accupel HDG 3000. Having no reason to suspect anything would be out of the ordinary I put up the 10 step split bars grayscale pattern from the Accupel and went to work on Expert 2, as Expert 1 had been set up on some settings from the net by the owner, Roy, and was looking not half bad already.
After several hours of aligning out the many non-linearities and tracking bumps of purple and green and other shades and getting that pattern to silky smooth grayscale tracking, it was time to come out of the working-on mode and see what real world material looked like. Before doing so, however, I knew there was one more thing that needed to be done. I could not see the registers in between the 10 IRE bars as I did not have a 20 step pattern, so once done with the 10IRE bars I knew the 45IRE, 55IRE, 65IRE bars etc. had not been touched. I knew their numbers would have to be extrapolated - set at the midpoint between each of the 10 step values that bordered them, above and below - so before I was through I set each of those accordingly. Where the 10 IRE bar for 40IRE was set for -10 and the 50IRE bar was set for -5, I knew the register for 45IRE would need to ideally be set for -7.5. So I set it for either -7 or -8, since no half steps were offered. It was not really going to matter which, -7 or -8, when we had all the way from +50 clicks up to -50 clicks down from zero point to choose from.
But to not
do this extrapolation would leave a setting of 0 at all 5 point registers. 50IRE would be -5 while 45IRE would be factory default of zero, and 40IRE would be -10. 80IRE would be maybe +15IRE, 85IRE would be zero, 90IRE would be +10 (not real world numbers, of course, just examples). Not very linear, many mini bumps to it, all the way up and down the IRE scale, invisible on the bars pattern but fully plugged in on real world content. So I corrected all those plus and minus 5IRE half steps in between the 10IRE steps, such that the invisible 85IRE register would be +13 and the invisible 45IRE would be -7, etc. This completed the smoothing out and linearity-making of the grayscale tracking, which wound up about as good as it gets. "Silky smooth", as the Yohan would say...
This was after having to constantly find a way to restore the picture's light levels, since with a fixed image up there they had been designed to protect the screen by dimming, ever so imperceptibly minute by minute, to prevent screenburn. On the 9800 we only had to click on Mute
. On the 9300 that was not enough, we had to actually leave the working-on menu and come back in, to restore the full light level.
I clicked on the pattern outside of the graphics and watched them all disappear, leaving just the pattern, as expected. Now just the split bars pattern from the Accupel remained, as if we were watching real world material. We were back to normal operation again.
But back on the pattern itself something just didn't look right. The corrections seemed to have not "taken" properly. There were a whole new set of bars that were halfway as off from each other as when I had started, but not in the same ways nor in the same directions. It was scratch our head time.
It took a while and more than a few tries, to finally nail it down. On full field window patterns I might not have gotten to the bottom of it at all. But 10 IRE step bars was perfect.
Whatever corrections I had made on one bar took fine while I was in the working-on mode. But the effect of the corrections shifted one bar - or +10IRE - to the right as soon as I left that mode and the set went back to normal.
It was totally obvious with the 10 step bars pattern. All we had to do was crank one register up to the max of 50 clicks UP on one color only - we used blue - and come out again, and bingo! The register 10 IRE to the right had taken the hit. On 40IRE the bar affected was 50IRE. On 70IRE, the one that actually took the corrections
was the 80IRE bar. Since I knew every step of the 10IRE steps given was going to be a little different from the next, this was a major find. As each bar required slightly different tweaking of each of the 3 colors involved, this was going to affect the grayscale's tracking linearity all over the place.
We couldn't believe what we were seeing! We tried it again and again and again, on every register in there. Every single 10 IRE register in there did the same thing. On coming out of working-on mode to normal, the bar to the right
had taken the corrections instead of the bar actually corrected!
So fellow calibrators, be ready for this little present being left at your doorstep here by LG's software. We tried it on the 9300 first and then the 9800. Both did exactly the same thing. We did whatever updates were called for, no change. Whatever corrections you make in one 10IRE register will "take" on the register one 10IRE step higher.
Who knows what's going to happen on current calibrations if they fix this gliche in an upcoming firmware update!