[argyllcms] Re: Pioneer Kuro Plasma calibration with dispcal
- From: Fatcat <molotok@xxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Tue, 14 Jan 2014 03:14:51 +0400
Hi Graeme and thank you very much for the time you took to answer me.
>> I'm having a big trouble calibrating the Kuro PDP-LX5090H plasma with
>> dispcal(gui) and
>> i1 Pro (ES-1000) spectrometer. The problem is that the results are extremely
>> inconsistent
>> even when the profile quality is set to "high" and the speed is set to the
>> slowest.
>
> Hi,
> I'm not sure quite what you mean by "speed set to slowest". Certainly
> you get more accurate low light readings when the instrument uses adaptive
> measurement mode.
There are two settings in displaycalgui, one of them is called "calibration
speed", that's what I was referring to.
> Plasmas's have a number of peculiarities, ABL being one of the major ones,
> which is usually dealt with by keeping the test square small enough not
> to trip it. (ie. 10% area).
This is very interesting I wasn't aware of ABL and I'm still not sure if that's
the case with this Kuro display. The issue is that it's enough to put a patch
of 100% white somewhere and the rest of the screen space in horizontal
dimension will darken somewhat. Perhaps this line bleed effect is compounded by
ABL. However, I did most of the attempts with the smallest size patch window I
could get from dispcalgui. Certainly much less than 10% of the area.
> Rather than continue to use dispcal and/or dispread over and over, I'd
> recommend
> just getting to the bottom of the reading consistency issue first. Once that
> is
> under control, the other tools should work as per normal.
>
> The tools I'd recommend for that are "dispwin -m" to display 100%, 75%, 50%,
> 25%, 12.5%
> & 0% grey, and "spotread -V" to check reading to reading consistency +
> standard deviation. You use the interactive "r" command to set a reference,
> and
> then you can check the delta E + statistics for subsequent readings. If
> you're getting
> delta E's > 1, then this points to a consistency issue of either the display
> and/or
> the measurement.
>
> You need to figure out where this is happening or where it is worst - near
> white ?
> In the middle ? Near black ?
And that was what I did. The results were extremely surprising. I did get
consistent readings within dE<1.0 for every primary color except blue! Pure red
was the best and showed very little error, then came the green which was a
little bit higher but still within 1.0 and pure blue was going anywhere between
zero and 4.3. It therefore translated to quite bad 100% white performance and
75% white as well, while 50% white and less were good. Same thing happened with
50% blue: the error margin was acceptably low. What does that mean? I certainly
didn't imagine anything like that happening.
I played with every available setting there was, reducing and increasing the
brightness of blue channel using display controls, doing the same thing with
brightness and contrast controls, it didn't change much. Perhaps in some
extreme cases dE 4 stopped appearing and the maximum was only 3.
Next thing I tried after returning everything back to normal was using
different brightness of the background. Curiously I got the best results using
the 100% white background. The results were still unacceptable for white and
blue but the error margin was less. Black background gave me the worst
performance and errors going up to dE 5 on a 100% blue patch. I tried using the
high resolution mode and standard one, didn't make any immediately noticeable
difference.
100% black was fine.
One thing I'm not sure about was how dE didn't change much when I switched from
pure white to pure black on subsequent measurements. Sorry about my ignorance
on the matter, but I thought that perceptually different colors should give me
a high error delta, but that wasn't the case. I compared black to white and red
to green and didn't get dE 150, just about "1" (without resetting reference
point of course). Is it because I was doing something wrong or is it how it
should be?
One other thing of note that I forgot to mention is that the original
calibration plate for the device has been lost and whenever I get a
self-calibration prompt I just press the opening to a high quality unprocessed
patch of a photographic paper placed on a table. Don't know if it matters much.
And finally all that might explain how I got better results with dispcal
initially: my background was changing every hour at first and those were mostly
very bright photos. Later I changed the background to a solid medium-dark gray
and the measurements started to be very inconsistent from one try to another.
Zakhar
Other related posts:
