Originally Posted by bobof
Does it work well in practice?
That depends on one's expectation. I'm sure a re-calibration wil always improve the accuracy further.
2) Similar to 1) - this surely only holds for regions of the gamut the autocal measures and fixes. It doesn't take long so I guess it doesn't do very much. Do people get very similar results profiling the projector across the whole gamut after subsequent autocals?
In the context of the current discussion, we're comparing auto calibration with manual calibration. For grey scale, the current JVC projectors actually have 256 points for each colour, way more than what's accessible from the manual calibration interface (12 points). I have no knowledge of the internal CMS, but keep in mind that manual calibration is limited to a single point for each primary colour.
3) Say the JVC aims for something like 1dE difference (I don't know, but I doubt it targets 0dE accuracy). This could be in any direction. If you apply a manual cal over the top. So your calibration may be further away than it was if the correction has moved in the "wrong" direction.
In my experience, autocal (using Spyder5) shows some residual errors when checked manually (using i1D3), but the errors are consistent (repeatable) and not random, which is not surprising given that the same algorithm is applied every time.
4) The 3 projectors I've used have exhibited colour temperature shift on changing aperture settings, which would be necessary if you wanted to maintain reference brightness. Can you do the JVC autocal at a particular aperture so this is calibrated out?
The JVC autocal stores independent colour temperature settings for 4 different aperture ranges. You can make 4 different calibration runs at different aperture settings (e.g., 0, 5, 10, 15) if you want to cover all apertures.