On Jan 14, 2014, at 10:07 AM, Fatcat <molotok@xxxxxxxxx> wrote: >> Hmm. That doesn't really explain the inconsistent black measurement values - >> they were going up and down together. > > Yes indeed. I was being dumb again. Between different colors there had to be > a huge delta, more than "1". It turned out I didn't put the tool into > emissive mode with spotread -V -e and it was shining its light onto the > display every time. I didn't see it as it was on a plate but you get the idea. > > Anyhow, now that it's been sorted out, I repeated the measurements and still > got weird results. Lighter patches behaved well. Problems started when I got > to 12.5% gray. First few measurements if taken right one after another fall > into dE<1, however each 7 seconds passed bump up the delta by about 0.15. So > in about a minute the error is 1 and growing. > > Take a look at this log. Most of the measurements were done in approx 7 > second intervals, for a couple others I waited 30 seconds or more. > Was the pattern on the screen the whole time or was it black in between measurements? > > The value is constantly drifting. More so in the luminance channel it seems. > > Several hours later I repeated the measurements and got the same pattern, the > values were constantly drifting albeit slower. On the same 12.5% patch it > took me about 4 minutes to get to dE 1. What's interesting is that between > subsequent measurements the difference usually doesn't exceed dE 0.3. > > The drift is faster on dark patches while on lighter ones it's much slower > although still present. On some patches it takes about half a minute for the > plasma to stabilize, i.e. you change the color, immediately take a > measurement, set it as a reference point and the value starts immediately > drifting, after 30 seconds the error is about 0.3 and from then on increases > slowly. On a cyan patch this initial drift was only present in L and b > channels, but absent in a. > > I can't interpret these results at all. What I understand is that given the > small error between measurements for me it probably would be better to use a > mode with very few crosschecks but instead pausing after changing color and > before taking the actual measurement. It's interesting to know your thoughts > on the matter. > > Zakhar The drift looks to be too large to blame on probe dark current drift so that would imply something odd with the panel. I have seen “fatigue” on plasma panels before but that goes in the opposite direction and is typical of bright patterns not dark. Can you repeat the measurements displaying the window on top of a non-black background (say 20% level or so), if the behavior changes it might be an indication of some dynamic contrast type of processing although this is usually done on much shorter time scales. The other possibility is that due to panel aging the phosphor efficiencies have declined and that constant stimulation allows some small amount of recovery. If that is the case you would have to keep them stimulated in between measures to stabilize. regards, -scott