Milan Knížek wrote: > x "fakeread values are normalised to 100" > > x meaning that dispread values are not normalised to 100 Hi, actually with the current code dispread can return either normalized (default) or non-normalized values (-w flag). This is marked in the resulting .ti3 by the NORMALIZED_TO_Y_100 YES/NO value. > My confusion stems from the test.ti3 (produced by dispread), which reads > that the numbers _are_ normalised: > > KEYWORD "LUMINANCE_XYZ_CDM2" > LUMINANCE_XYZ_CDM2 "66.758160 61.640342 95.172183" > KEYWORD "NORMALIZED_TO_Y_100" > NORMALIZED_TO_Y_100 "YES" ccxmake will take either case and restore the absolute values for computing the correction matrix. Since the fakeread output doesn't have the information needed for this to happen, you would have to add this, by adding the LUMINANCE_XYZ_CDM2 value and NORMALIZED_TO_Y_100 "YES" so that it too can be converted back to absolute by ccxxmake. > Is your hint re. copying of LUMINANCE_XYZ_CDM2 values from test.ti3 to > ref.ti3 still valid nowadays? Yes, since I haven't added the code to fakeread to read the ICC SigLuminanceTag and use that to set a LUMINANCE_XYZ_CDM2 value, even if it is present in the profile. > BTW, my vendor tailored Huey for Samsung SyncMaster XL20 seems broken > (all of sudden there is too much green after calibration), hence I had a > similar idea as the original poster to re-use my Eye-one Display 2 and > rely on Samsung's emulated sRGB to create the correction matrix. Hmm. Either something has gone wrong with it electrically, or the filters have aged. Graeme Gill.