"I wonder if a company like EIZO already does this sort of thing." N.C.
Indeed, a process similar to what you describe is done in Eizo monitors. It would have to be done by any monitor manufacturer if high spatial uniformity of output characteristics is needed. That is, all LCD panels are manufactured with nonuniformities derived from several steps in the complicated process. On top of that all back lights add an another layer of spatial variation. The end result is that spatial correction is mandatory today for high quality performance.
Measuring and correcting for each pixel is not done, nor does it have to be in most cases. One result of piling up nonuniformities from different sources, some of which may offset others, is that the is that the cumulative nonuniformity is slowly varying (spatially). So the screen is conceptually divided in an array of zones and those zones are optically measured (automated) and corrections are derived for each zone and all pixels within that zone receive the same correction via hardware calculations & LUTs. These corrections and LUTs are separate from, and follow in the data stream, those applied to the RGB input data to achieve color balance, accurate gamma and smooth tone transitions. SO there is relatively high performance silicon and 14-bit calculations going on inside an Eizo monitor and that plus the time and capital to physically characterize the front-of-screen performance is why the prices are what they are. Also, Eizo starts with a tighter-than-typical spec on the raw panel and that also drives up the price.
Good post Paolo. I know that Eizo has a program characterizing the entire monitor starting from screen selection, and you did a nice job explaining it.
The zone array and my idea of treating a set of pixels derived from circle of confusion concepts sort of match up. I didn't think that it would actually go pixel to pixel.
Do you know the size of the zone?
I' d bet Eizo doesn't scrap the panels that don't meet expectations. Probably sells them to Dell!
"Do you know the size of the zone?" - H.chrome
No I do not. That would be inside information as I've not seen it published. Related correction schemes have used a uniform array of zones, something like an 8 by 8 array, but it is possible and perhaps even desirable to have a nonuniform array of zones. That is, match (in an approximate way) the corrections zones to the measured nonuniformity.