[argyllcms] Re: fwa compensation unexpected results

Roberto Michelena wrote: 
I got a proofing paper heavy in FWA content. I measured it in my
DTP-70 without UV filter, it has a b=-8.8. With UV filter it's reduced
to -3

So I tried to use Argyll's algorithms to correct the non-filtered
measurements, in order to compare with the UV filtered version.
Unexpectedly enough, the FWA-corrected ti3 file yields a b=-11.3 value!!


You have a omitted a critical piece of information :
what viewing illuminant did you tell profile you were going to use ?

By default, profile assumes D50. Not D50 white point, but Daylight
spectrum, 5000 kelvin.

Consider: the instrument uses a tungsten light source, 'A' spectrum.
It has a moderate level of relative UV in it, and gives you a b = -8.8.
D50 has a slightly higher level of relative UV, so the FWA will get more
stimulation, and will get more blue. So b = -11.3 sounds like it's working
just as expected.

Now, if in fact you are not viewing in actual Daylight 5000K light,
and your illuminant has a lower relative level of UV than the 'A'
spectrum in the instrument, then the resulting print will look less
blue than the profile anticipates it will.

So the answer is, for FWA compensation to work properly,
you need to specify your viewing illuminant spectrum. If you're
doing proofing, the FWA compensation needs to be applied to
both the input and output profiles, and the viewing illuminant spectrum
you want them to match each other under, must be specified when making
both profiles.

Now the real difficulty is in getting a spectrum for your viewing
illuminant. Many people use D50 simulators, often of the fluorescent
kind. These have widely varying levels of relative UV. Simple ones,
have even less UV than a tungsten light, because the phosphor coating
in the lamp, soaks it all up. Some high end lamps have extra medium
wavelength UV phosphors in them, so that they have a much higher level
of UV in their output (to better match actual Daylight - though sometimes
I think they may have gone too far).

Ideally you want to measure your illuminant. (You can use
spotread -e -s, and cut and paste the resulting values into
an Argyll .sp file. I may fix this to be more elegant at
some stage).

The catch is that all the popular graphic arts spectrometers
don't measure to short enough wavelengths. The FWA is sensitive
down to something like 300nm, but most instruments cut off at
380 or 400nm. You can try extrapolating, and adjust your
extrapolations in light of the results you get. Remember
that changes will affect the input and output profiles in
opposite directions though, and whether you have to increase
your extrapolation or decrease it will depend on whether your
input media or output media has the greater level of FWA.

Another approach is simply to try some of the stock CIE illuminants
that I've put into Argyll - some are characterized down to 300nm.
You've got A, D50, D65, F5, F8 and F10 to choose from. The 'F' series
are fluorescent lights, and F8 & 10 are 5000 degree white point. The
CIE specification assumes that the 'F' series have 0 output below 380 nm,
which may or may not be true.

In the 'ref' directory I've put a few more spectra  (.sp) that have
been measured and extrapolated (and in some cases tweaked).
I can't really vouch for how accurate they are. [What I'd
really like is an Ocean Optics spectrometer module,
configured to go down to 300nm :-) ]

Hope this helps,
        Graeme Gill.

Other related posts: