[argyllcms] Re: i1 Display bundled with Samsung XL20

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 05 Apr 2009 22:58:37 +0200

Graeme Gill wrote:
>> Display, or whether the eeprom possibly contains a calibration matrix
>> especially tuned for the LED light source of the XL20? (there is at
>> least a big inscription "LED SyncMaster" on the body of the i1)
>> [...]
>> expected). Does somebody happen to know, whether this is a "regular" i1
> My guess from an engineering point of view would be that it contains
> a standard calibration, and that any special adjustment done to suite
> the display would be in the software that accompanies it,

My XL20 was unfortunately ligering around for a while (while I was busy
meeting other obligations), but now I found a little bit of time to dig
it out again and to do a few experiments. I'ts now also the first time I
got the "Natural Color Expert" working under Windows XP [last time when
I tried (on a different PC), the NCE did not find the monitor - I guess
it was an issue with the graphics card, when accesing the I2C bus]. So I
had the opportunity now to compare the measurements from NCE with
measurements from Argyll, and they are virtually identical [within less
than +/- 0.0005 chromaticity units (NCE only reports 3 decimal digits)].
This obviously implies that very likely my "LED Syncmaster i1 Display"
does not contain the standard LCD calibration, but rather an LCD
calibration matrix for the XL20, and it's obviously not the NCE, which
performs any special adjustments for the XL20.

I was also surprised about NCE's calibration procedure. First (as
expected) it displays a white patch and one can see that it adjusts the
backlight to meet the selected white point target, but then it measures
only the three primaries and black (black is likely adjusted too,
according to the selected black target; but I had selected "mininum", so
I did not notice any change of the displayed black patch). However, in
order to establish the calibration, NCE does NOT measure a gray ramp, as
dispcal does. And eventually [as I was afraid after this observation],
when I ran dispcal on the NCE-calibrated XL20 it turned out that the
in-monitor calibration established by NCE is not perfect, it does not
meet the gamma target [measured about 2.0, rather than 2.2 as I had
specified in the NCE GUI], and the chromaticity of the R=G=B ramp is
also still a little bit off neutral too.

I also did take a brief look with IDA at some DLLs coming with the NCE.
I found a couple of I2C_...() functions, and I think they are the
central point for doing all the DDC/CI I/O. So it might be interesting
to trace all invocations of these functions. I tried to run NCE under
GDB and to place breakpoints on these functions, however it looks like
the protocol is rather timing-sensitive and does not like too long
delays (as a result, strange, magic adjustments happened on my display,
while NCE was running under the debugger, and evetually NCE even locked up).


> but one way to tell would be to dump the calibration matrix by turning
> on #define DEBUG in i1disp.c, and comparing it to another device.
> For my (non-XL20) i1display I get:
> LCD/user cal[0] = 0.092167
> LCD/user cal[1] = -0.003769
> LCD/user cal[2] = 0.024447
> LCD/user cal[3] = 0.001060
> LCD/user cal[4] = 0.060454
> LCD/user cal[5] = 0.000752
> LCD/user cal[6] = -0.000367
> LCD/user cal[7] = 0.000713
> LCD/user cal[8] = 0.161666

Other related posts: