[argyllcms] Re: Using Argyll with an old densitometer (non-XYZ)
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Wed, 22 Mar 2006 00:40:23 +1100
Thomas Kumlehn wrote:
Did any of you already use an X-Rite 810 colour densitometer, just capable
of reading Neg densities in RGB ?
What you're asking for can be done, but it needs a fair bit of work,
if you have to do it all yourself.
Basically you have to create a mapping between the negative densities,
and CIE values of whatever it is you think you want to control.
At one extreme this might be the colors that the end viewer sees,
after the image on the negative has gone through the complete
production chain, or you may pick some intermediate point,
and attempt to control it instead.
The bottom line though, is that you are going to have to measure
some test patches with both you X-810, and a colorimeter/spectrometer,
in order to create the mapping.
I think some of the folks involved with the ICC Motion Picture Working
group have attempted this type of thing (see
<http://www.color.org/groups.html#digital2>).
[I know at least one of the people characterizing film built an automatic
film patch reader using a Gretag i1 out of LEGO, so that they could read their
whole
33^3 grid of motion picture film test patches automatically.]
There are probably some articles out there that could help you.
See <http://www.imaging.org/store/epub.cfm?abstrid=22202>,
<http://www.imaging.org/store/epub.cfm?abstrid=22204>,
<http://www.imaging.org/store/epub.cfm?abstrid=22205>,
<http://www.imaging.org/store/epub.cfm?abstrid=6893>
for instance.
Or can anyone point me to a book or web source how to convert the X-810
RGB-densities to CIE-XYZ space as needed for CMS. I need the matrix
coefficients or the primaries vectors for the color filters and light
source used in the X-810.
This isn't going to take you in a useful direction. The filters and light
source in the X-810 are irrelevant, since they are optimized for
process control, not reading CIE values. The mapping between density and CIE
values could possibly be modelled mathematically (after taking
a lot of measurements to create the models from), but in practice you
might be better off created a general (LUT based) transform based
on a large set of measurement pairs. This is analogous to creating
a device profile, except that instead of device values, you
use density values. After creating the "profile", you can then
map three densities to a CIE value. Such a transformation will
be specific to certain film stock and production chain of course.
If successful and of interest for this community, I'd be happy to provide
the sources for my filmread solution. But regarding the reduced
functionality of the X-810 compared to the newer devices, I doubt it'll fit
to the general device interface used by argyll.
The good thing about it - used X-810's are availlable and much cheaper than
color spectrometers.
But cannot substitute for a colorimeter or spectrometer.
Graeme Gill.
Other related posts:
