Raphael Weniger wrote:
Besides a 3x3 matrix I also thought about some kind of 3D-curve to correct possible non-linear characteristics of the display/colorimeter. But probably there comes a point were you just should get a spectrometer instead of correcting a colorimeter.
I don't think that would be easy, and there is no indication that the popular spectrometers have noticeably better linearity than the colorimeters.
I couldn't find any sample files to try out the matrices I already have so I build the following one with the help of the source code and other files (.cal, .ti1/3):
OK, there are some sample files now: <http://www.argyllcms.com/doc/ccmxs.html>. Graeme Gill.