Let me correct some misinformation.
- The OEM model does NOT have a "fixed exposure time". This confuses the OEM i1 Display 3 Pro with the ColorMunki Display, which is a less expensive and less capable version of this meter.
- The video at this link http://www.chromapure.com//bugtest/10/10percent.html is not the video the poster is talking about despite being the link referenced. This video, as the path indicates, was one I uploaded to demonstrate some reported issue with a bug the details of which I have forgotten. The poster is referring to this video http://www.chromapure.com/movies/Display3/RGB/rgb.html that was posted 17 months ago as part of the initial release of support for this meter.
- The Display 3 or Display 3 PRO has no issues with repeatability down to very low light levels under the conditions described. This is for two reasons. First, since initial release, we have continued to refine and improve support for the meter based on the experience gained, and this has improved performance. Second, even if that were not the case, the video leaves a misleading impression because of the type of averaging we use in continuous mode, which samples many readings before reporting a result. The sample taken is larger the lower the reading and the read time is also slower at very low light levels. Taken together this means is that the 10% reading was initially outputting values that were still influenced by the 20% reading taken immediately before. This continued for several seconds until all of the 20% data was purged from the data set leaving only 10% readings to contribute to the reported value. This is clearly not a repeatability problem, which would show itself as random fluctuations. It is a slow and steady reduction in the amount of red reported, because at 10% gray the projector was actually putting out less red than at 20%.
To demonstrate this, I just captured another video.
See that the first time 10% is measured, it takes several seconds for the xy values to stabilize. The readings start at x0.309, y0.329 (R96, G101, B101) and then stabilize at x0.314, y0.333 (R99, G101, B97). This is a 4% change unlike the 13% change seen at 10% in the July, 2011 video. The difference is largely due to the improved support I mentioned. The remaining is due to the lingering effects from the previous reading. Again, this is not a repeatability issue. It is just the time lag the system requires to purge the data from the 20% reading. When I close the module and then reopen, thereby purging all of the data, and then measure 10% again, it goes to x0.314, y0.333 (R99, G101, B97) immediately with virtually no fluctuation. You'll also notice that when continuous readings start for the first time at 30%, it takes about 5 seconds for them to settle down. This is normal.
I also took single readings at 20% and 10%, which eliminates the averaging present in continuous readings, using a flat panel. This is probably the best test of repeatability.
Regarding speed, I posted an updated video on this as well. The Display 3 is a little slower in ChromaPure than it was at initial release, again because of our refined support, but it is still quite fast, completing a 10% grayscale run in about 13 seconds.
Finally, this performance is what you see for projectors and LCD flat panels. It is not quite as good with plasmas. To get good repeatability taking single readings on a plasma below 20% you really need to use Measurement Smoothing. Continuous readings perform about the same--the built-in averaging smoothes out the rough edges. Also, plasma read times are a little slower. The grayscale run takes about 22 seconds.