Holger Kiemes wrote:
I allready read the "Description of the .ti3 format" but I have still some questions regardig the data sets (the lines between BEGIN_DATA and END_DATA): I am want to create only (s)RGB profiles. 1) which values are written in a dataset? is it the expected color value in RGB and the measured value in XYZ?
I'm not quite clear on your question. The important information in a .ti3 file is each patch connection between the device value used to print the patch (RGB), and the color value measured for it (XYZ or LAB value). The index is not so important. An "expected" color value is something present in .ti1 and .ti2 files, and is used to aid the chart reading.
2) if there are more than two value pairs in a data set, are these severel measured values to be averaged by profile or are this different formats of the one measured value (like XYZ and L*a*b*)?
I'm guessing that you're asking what happens when there is more than one patch with the same RGB value ? This depends on what is processing the .ti3 file, but for colprof, the values are effectively averaged by the regular spline. One way of thinking about it is that two values with the same RGB are averaged, with the result being given twice the weight. This has subtle implications for the level of smoothing. Increasing the weight of the data points is the opposite of increasing the smoothing. Note that this "averaging" is what's going on even if the test points aren't exactly coincident.
3) Is the order of the values in the dataset importent, do i need to switch RGB, XYZ values to XYZ, RGB according to the value of COLOR_REP?
The order of the values within each line nee is important, and needs to correspond with the declaration in COLOR_REP. The order between the lines is not important, and the index value will be used if you're comparing sets, say using verify. Graeme Gill.