As Graeme pointed out, the choice of a single parameter for channel linearization is limited... at least when the same parameter is used for all inks. And Chroma is not a particularly good one either. The 'dE to paper' method I'm not sure, since it rarely really evens out, at least on inkjets... it does diminish the change rate (1st derivative), but still, where to cut? It seems to me that CIE X,Y,Z are the best choices; extend better than densities (do not level off so quickly), are much less noisy... X for Cyan, Y for Magenta, Z for Yellow, and any of them for Black. Anyway; I did a comprehensive Excel spreasheet full of graphs of many options, and pasted in it the measurement of a 256-step per channel wedge (1024 patches) from an Epson 4800, driven by an ORIS rip raw (without any linearization or color management). The only caveat is that there's a fixed mixing of light inks done by the RIP... this would be much more realistic if done in pure CMYK inks (a 4-color printer or a RIP that can ignore the light inks). I've posted the spreadsheet (3.2Mb) at: http://www.mytempdir.com/1109662 -- Roberto Michelena