Dear Klaus, >> Suppose I need to check a printer profile for reproduction of a set >> of about 500 pre-determined colours; colours being defined as XYZ D50 >> 2 degree. >> >> As a first step, I filter off the colours that are out of gamut >> using xicclu -fif -ia -pX > > This is a pragmatic way to check if the colors are in gamut. I tried out > others, but came back to it as it is quick and easy. > > Please note that you have to use -fb (lower precision) or specify the > original separation parameters if you want to use them with xicclu -fif. > Otherwise you use xicclu's defaults (-kr, estimated TIL, no black ink > limitation). Separation parameters may affect the gamut, especially in dark > areas. Yes, of course, separation parameters are to be included for CMYK printers. > >> Next, for those XYZ that fall into the gamut I convert them to RGB >> xicclu -fb -ip -pX > > Why perceptual? Do you want to investigate gamut mapping? I actually run it with all 3 intents to have an idea of mappings. As the list is already inside the gamut results from comparing runs with different intents are quite interesting to me. -p resulted in the least deltaE 2000. > >> From the RGB list I >> make a target formatting to CGATS and using ColorPort, print it, >> measure it, and compare resulting XYZ numbers to source XYZ. > > I'm not quite sure what you are after. If it's just the profile you want to > test, I'd skip the printing and measurement part and just complete ther > "round trip" with another xicclu call. Otherwise you bring additional > unknowns into play (printer repeatability, measurement uncertainty) For this case I'm after the check how the actual printer behaves. Doing several runs of printing and measuring I see deltaE between runs peaking at 5 with 3 being an average. The task is to evaluate a prototype of a new printing system. All usual visual checks of this system profiled using Argyll are very good > If you really want to test the complete workflow, you can of course also use > Argyll's tools (printtarg and chartread) to create and measure the target. The problem with using chartread here is that it does not support neither my iCColor nor Barbieri. -- Iliah Borg ib@xxxxxxxxxxx