[argyllcms] FWA overcompensation bug in spec2cie

  • From: Alexey Gribunin <Gribunin@xxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Wed, 14 Jul 2010 14:16:09 +0400

Hello Graeme!

It seems to me that I found bug in spec2cie.

It's a FWA overcompensation of LAB (and XYZ) values.

I'm running the following command on Test.ti3 measurements, which contains FWA (paper Lab values are 95.71 -0.48 -2.91):

spec2cie.exe -i ref/D50_0.0.sp -f -o 1931_2 Test.ti3 Test_UVCut.ti3 #command 1

In Test_UVCut.ti3 I have both spectral and LAB values. I think, they should match but they don't!

Spectral values seems to be FWA compensated correctly. After converting spectrals to Lab with the following command: spec2cie.exe -i D50 -o 1931_2 -n Test_UVCut.ti3 Test_UVCut_CIE.ti3 #command 2
I got paper Lab values 95.61 -1.33 0.24.
The same correct result I got by converting spectral values from Test_UVCut.ti3 to Lab in XRite MeasureTool (it's little bit tricky to extract spectral values from TI3 but it works).

But LAB values which I got in Test_UVCut.ti3 (#command 1) are wrong! Paper Lab values are 95.61 -2.37 2.64. Overcompensation! It seems to me that #command 1 makes FWA compensation twice. First time it makes compensation of spectral values, and the second time it makes compensation when converting compensated spectral values to Lab.

Best regards,
Alexey Gribunin,
UNIT Color Technologies,
Moscow, Russia.

Other related posts: