[argyllcms] Re: spec2cie

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 16 Sep 2013 15:20:09 +1000

Joe Tschudi wrote:
> Hello,
> I am getting weird results out of spec2cie when converting spectral data to 
> CIE data by
> using a custom illuminant. First I tried spotread (spotread -v -s  -c1 -a
> CustomTest_spotread.sp) to get ambient emission curves to be saved as an .sp 
> file. I
> also tried illumread (illumread -v -c1 CustomTest_illumread.sp) but the 
> results after
> converting with spec2cie are even further off. I have spectral remission data 
> that I
> measured using the same device (i1Pro 2) with chartread (chartread -v -c1 -p 
> -L -S
> TEST) which I am trying to convert from spectral data to LAB using the custom
> illuminant with spec2cie -v -n -i CustomTest_xxx.sp spectral.ti3 cie.ti3

        CustomTest_illumread.sp seems a little strange. specplot shows a CT of 
and an L*a*b* compared to D50 of 100.000000 -12.215330 -27.322065, which is
quite blue.
> Although the CCT of the emission curves shows me a value of around 5850K I am 
> getting
> LAB results that are completely blueish and have nothing to do anymore with 
> the
> original spectral color. To compare the results I am using Bruce Lindbloom's 
> Spectral
> Calculator Spreadsheets (10nm) (http://www.brucelindbloom.com -> Calc -> 
> Spectral
> Calculator Spreadsheets -> scroll down for 10nm Download) where it's possible 
> to fill
> one custom spectral (remission) column and a custom (emission) illuminant and 
> the
> results there are correct IMHO.
> Using regular spec2cie with -i D50 I get: 2° LAB 91.21 / 0.15 / 4.34
> With the Spreadsheet I get: 2° CustomIlluminant_spotread: LAB  91.12 / -0.47 
> / 4.49 2°
> CustomIlluminant_illumread: LAB  91.04 / -1.12 / 4.50
> With spec2cie I get: 2° CustomIlluminant_spotread: LAB 91.12 / -4.29 / -4.91 
> 2°

I'm not sure why you think that is in error. CustomIlluminant_spotread is bluer
than D50 (L*a*b* of 100.000000 -4.141052 -10.421183), so there is no
surprise that your samples illuminated by a bluer light has a bluer
XYZ value.

Beware of the difference between reflectance values and absolute XYZ values 

It's also not clear what the white reference is for the conversion to L*a*b* in
Bruce Lindbloom's spreadsheet. If it uses the custom illuminant as the white
reference rather than D50, then naturally you will get a more neutral set of
numbers to the spec2cie result.

Graeme Gill.

Other related posts: