Color of D50*.sp?

[Ben Goren]

I took a couple of the D50*.sp files in the ref folder, minimally munged them into .ti3 files, and ran them through spec2cie. I was expecting to get Lab = 100 0 0 (within rounding), but the results are rather different.

For the one with no UV, it gives X Y Z L a b: 91.6017 98.3876 65.4797 99.3732 -5.76953 13.7374

For the one with 100% UV: 92.2926 98.4088 68.7988 99.3815 -4.57256 10.6742

Am I missing or misunderstanding something?

Thanks,

b&

[Graeme Gill]

Post by Ben Goren
I took a couple of the D50*.sp files in the ref folder, minimally munged them into .ti3
files, and ran them through spec2cie. I was expecting to get Lab = 100 0 0 (within
rounding), but the results are rather different.
For the one with no UV, it gives X Y Z L a b: 91.6017 98.3876 65.4797 99.3732 -5.76953
13.7374
For the one with 100% UV: 92.2926 98.4088 68.7988 99.3815 -4.57256 10.6742
Am I missing or misunderstanding something?

Hello Ben,
my guess would be that you converted it to a .ti3 *reflectance* spectrum.
If you put that through spec2cie then you are effectively squaring the
D50 spectrum, and you would end up with something other than what you were
expecting.

Instead you should convert to a .ti3 emission spectrum.

Note that specplot provides a direct way of checking the color of an emission spectrum (.sp).

Note that due to the calculation methods used, the tabulated D50 XYZ value that is
used as the ICC reference will not quite correspond to the XYZ value computed from
the D50 5nm spectrum created by the Daylight spectrum interpolation formula.

Graeme Gill.

[Ben Goren]

Post by Graeme Gill
my guess would be that you converted it to a .ti3 *reflectance* spectrum.

You're absolutely right.

And I'm very glad I asked, because that's not something I was even aware of...and something I likely would have gotten bitten very badly by on a current project had I not just stumbled on it and you pointed me in the right direction.

Thanks!

...if all goes as planned (Ha!) I'll soon be able to repay you with a new, and hopefully superior, method of camera profiling....

Cheers,

b&