On Sun, 2008-12-14 at 21:14 +0000, Alastair M. Robinson wrote: > Hi :) > > Leonard Evens wrote: > > > There is no question that the calibration information is being > > loaded. It is what the applications are doing with the profile that > > seems to be questionable. > > Or what the profile itself is describing. A monitor profile is only > valid for the calibration state for which it was produced - and from > what you said about the image on screen looking like it's been > overcorrected, I suspect your profile has been built to describe the > monitor in an uncalibrated state. > > The Calibration step is, in fact, optional - it's perfectly possible to > build a profile to describe the monitor in any state of calibration - > though the results are generally better after calibration. > > To try and pin this down a little further, can you try the following: > Display a greyscale image of some kind ramp in an un-colour-managed > application, then do this: > dispwin -c > (resets the LUTs with a linear ramp) > Do you see a colour cast to the ramp? Give your eyes a few moments to > adapt before deciding! > > Then: > dispwin -L > (Loads your profile, assuming you've installed it with dispwin -I at > some point.) > > Do you see a change in the display? > Do you now see a different colour-cast in the greyscale, Yes. > or is it now It is closer to correct. > correct? > > If you see no change, can you try this: > dispwin /path/to/file.cal > (Load the LUTs using the .cal file created by dispcal.) There is no discernible change between dispwin -l and dispwin xxxx.cal where xxxx.cal is the calibration file with which the profile was created. In this case I used the targen, dispread, colprof sequence to produce the profile, and I definitely used the -k option. Perhaps I should remark that I am now up to W in my lettering having started on this occasion with D. Last year I got to C and stopped there with what seemed an acceptable calibration/profile at the time. So there has been no want of trying.) > > If the effect of this is any different from dispwin -L then your profile > doesn't correctly contain the LUT data - the most likely cause of this > being the omission of the -k parameter when creating it. > > Another thing to try: > dispwin -L > (to set the X monitor profile atom) > xgamma -gamma 1.0 > (to load the LUTs with a linear ramp - could also use dispwin -c for this) > Now display a greyscale ramp in a Colour-managed application. Do you > still see a colour cast? It is actually worse that way. > > Hope this is some help, > > All the best, > -- > Alastair M. Robinson Well, these were interesting things to try. I am beginning to fear that my problem will only be resolved after I completely understand in the goriest detail exactly what all these programs are doing and how they relate to what I see. I also wonder if somehow the problem is me. When I had cataracts, everything had a slight yellow cast, but my cataracts were removed in 2001. Also, I remember when I did color printing in my darkroom, 25 years ago, I had great difficulty getting rid of color casts in gray scales, despite using an expensive color analyzers. I seemed to be specially sensitive to very small color shifts, as small a .025 in color filter units (if I remember the scale correctly). It could also have something to do with ambient light. -- Leonard Evens <len@xxxxxxxxxxxxxxxxxxxxx> Mathematics Department, Northwestern University