[argyllcms] Re: FWA overcompensation bug in spec2cie

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 15 Jul 2010 12:03:25 +0200

Am 15.07.2010 11:19, schrieb Alexey Gribunin:
> Again, the problem happens here, in this command. Test_UVCut.ti3
> contains both spectral and LAB values and they are different.
> Spectrals are compensated with D50_0.0 only once, but Lab values two
> times because Lab calculated from spectrals with D50_0.0.sp again.

I think you are assuming that the FWA compensated spetral reflectances
would be generic. This is however not the case. If you do a FWA
compensation for say D50_0.0.sp, then the resulting compensated
reflectance spectra are only valid for D50_0.0.sp light. Under a
different light source (e.g. regular D50), the FWA compensated spectra
would look different as well. So I don't see why it should make sense to
compute FWA compensated spectra for D50_0.0.sp, but then assume a
regular D50 illuminant for converting these compensated reflectance
spectra to XYZ. Goal of the computation is to estimate which XYZ color
the samples would have, when they are illuminated with a particular
light source (specified with the -i option). So the same light source
spectrum needs to be used for computing a) the FWA compensated
reflectance spectra and b) the XYZ numbers from the compensated
reflectance spectra.


Other related posts: