I've just finished an analysis of the variability in i1iO measurements, which led to a suggested protocol for measuring printed matter with the X-Rite i1iO scanning table (see the write-up at
http://www.99main.com/~centore/ColourSciencePapers/i1iO2MeasuringProtocolLowRes.pdf ). The takeaway is that the 1i1O occasionally (on about 1 out of 10 pages) returns some significantly outlying measurements, which can adversely affect further processing. To overcome this problem, a measuring protocol is suggested, involving the median of nine separate measurements. The protocol is time-consuming, but effectively eliminates the outliers, and decreases the errors in an interpolation algorithm by 40 percent, even with half the input data. Despite its time demands, the protocol would likely be useful for profiling or reproduction algorithms, where measurements are made once and used for years afterward. Any comments, criticisms, questions, or suggestions are welcome.
