[argyllcms] Re: Using a RGB-Camera as colorimeter
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Sat, 4 Nov 2017 12:06:52 +1100
Charlie (Redacted sender chpine2 for DMARC) wrote:
The people who do Raw Therapee make .dcp profiles from ColorChecker shots
submitted to
them per their instructions.
The problem is that a camera is not a colorimeter - it's spectral response is
not the
same as a human. So even if a perfect mapping is created between one
combination of
3 primary reflectances (i.e. a ColorChecker) and CIE values, it will have
errors when
interpreting a different one (the printer output) as CIE color values.
(This is not a limitation when trying to make pretty pictures - you are stuck
with the fact that a camera is not a Colorimeter anyway, and have to make
the best of it. But photo's are generally not intended to be a scientific
recording of a scene.)
And this ignores a lot of other realities - inks do not mix together in a
perfectly additive way, and typically more than 3 inks are used anyway (Black
is very useful). So there can be no perfect mapping between ColorChecker value
and CIE values, or printer values and CIE values via the camera, even if you
were
to profile the camera using a chart made on the printer itself. (The latter is
a best case scenario, but is pointless because you need a colorimeter or
spectrometer
to measure the test chart as a reference.)
Cheers,
Graeme Gill.
Other related posts:
