[argyllcms] Re: camera profiling with newest version of Argyll

  • From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 27 Aug 2015 19:09:48 -0700

On Aug 27, 2015, at 6:32 PM, Graeme Gill <graeme@xxxxxxxxxxxxx> wrote:

The first patch has L* a* b* = 96.709 1.8667 -13.059, which is decidedly
bluish

I'm not sure where you get that value from - spec2cie says it is
96.694984 -0.854171 0.157998.

The ColorCheckerSB-munki.cie file he posted earlier has LAB_L LAB_A LAB_B as
the last three columns, and 96.709 1.8667 -13.059 for the first row. That's
what first screamed at me, "FWA!"

but the ColorMunki values below 420 are "guestimated", since the illuminant
drops off there, and the diffraction grating looses efficiency.

It's the values from 380 to 420 in the file that next screamed, "FWA," to me,
especially coupled with the peak at 480nm and the decline thereafter. And, of
course, having first been primed by the -13 b* in the file. But if 380 - 420
don't actually mean anything...then they don't actually mean anything. (In
which case I think it'd be better to not have them at all, but that's another
story....)

The A1 white patch D50 L*a*b* comes out at 97.222621 -0.954457 0.777881 for
the Spectrolino, and 96.694984 -0.854171 0.157998 for the Munki.

So I think you are on the wrong track.

Well, *something*'s not right with that .ti3 file. I just ran spec2cie on it,
myself, and compared the original XYZ and LAB values therein...and spec2cie
comes up with radically different results than what's in that file. To wit:

XYZ_X XYZ_Y XYZ_Z
Old 89.467 91.727 91.969 (ColorCheckerSB-munki.cie from Chad)
New 87.946 91.694 75.454 (Same file, after spec2cie)

LAB_L LAB_A LAB_B
Old 96.709 1.8667 -13.059 (ColorCheckerSB-munki.cie from Chad)
New 96.695 -0.8541 0.1580 (Same file, after spec2cie)

So, unless Chad was supplying arguments to colprof that would cause it to
re-calculate reference values from spectral data...those differences alone
would more than explain the troubles he's having.

I may be crazy, but I don't think I'm _entirely_ off the reservation...but, if
I am, I'd sure want to know about it....

I think I've learned one thing...and that's not to simply assume that XYZ and
LAB values in a file can be counted on as corresponding with the spectral data
they're accompanying -- or, at least, not the D50 calculation thereof. If I run
spec2cie with -i D65, the values (XYZ 86.68459 91.73454 99.42845) are closer to
the original but definitely not a match, so that's not it, either.

b&

Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

Other related posts: