Originally Posted by sotti
A dE of 3 for grayscale is adequate, but is definitely not optimal.
If you want to see the absolute best we can do, turn down the dE target to 0. There are a host of changes for AutoCal's algorithm so you will get the best result you've ever seen from CalMAN. Granted it usually only takes about 3 or 4 reads to get under a dE of 0.5, but it'll take 8-12 reads to get to 0.3. I can't say the trade off is woth it, but if you want that last click or two of accuracy 1092 can deliver it.
Yes I use a dE of 0.5, and have advised other to do the same.
Ok, Just ran the new Beta 5.1. I haven't viewed any thing yet, I just wanted to post when these things are fresh in my mind.
21 Point grayscale took 20 min to complete and about 180 reads. Normal is about 8 min and 60 reads. I have NP with this longer time and more reads.
The GS in the lower IRE's had about a dE of about 3. However after running the cube the "verify calibration" showed the GS was at 0.2 and the chart showed very little error.
Gamut luminance seem to have a little problem with Blue showing a error of about 8, all the rest where about 4 or less. But this is better than it has been with 5.0.4.
LUT cube took about 22 min to complete.
The "verify calibration" showed the colors hitting there mark pretty well.
So on paper it sure looks like the LUT cube workflow and my Mini 3D are starting to play nice. Also check to make sure the CMS and grayscale had been written to my Mini, all looks well there.
I will add to this post after viewing some video.
It looks like your RGB balance needs more tuning in the lower IRE's. My guess is with the Luma/Y factor.
The Gamut and luminance is realty nice for the LUT cube, but it almost looks as if the Gamut/luminance is being used to help reset the RGB/gamma balance.in the GS.
Im sure you (Joel) will be able to correct this in no time. Thanks.