Three-Dimensional Control of Color Management Systems - Control of Color Imaging Systems: Analysis and Design - page 374

FIGURE 7.45 (See color insert following page 428.) Hair image rendered with K-restricted
ICC profile whose inversion response is shown in Figure 7.44. (left: original; middle: simulated
using ICC profile from Example 7.8 with 3-to-3 control-based inversion; right: simulated with
K-restricted 4-to-3 control-based inversion without preconditioning [step (h)]).

The above procedure may take more computing to complete all the steps,
especially due to large time spent in step (c). Hence, a faster way to build the pro les

is to run the procedure in two groups. The first group can be used to build the initial

CMYK LUT and the gain matrix in the factory on a nominal print engine for all the
node colors (includes in-gamut and out-of-gamut node colors). The second group
can be used for running the iterative control algorithm of Figure 7.41 on a drifted
print engine in the

field. These initial CMYK and gain matrices can be carried as

source starting LUTs for the control algorithm. For this kind of architecture, it is

important to note that the initial CMYK and gain matrix values need to be found for
colors that the printer cannot render in its nominal state but are required to cover the
drifted printer gamut. This can be done by extrapolating the printer model for the
nominal state or through extrapolation of the initial CMYK values for the in-gamut
node colors produced on a printer under nominal state.
Apart from the speed advantage,

=

there are other bene ts in carrying the

GCR-constrained initial CMYK and gain matrix values to the eld. We can incorporate

different UCR approaches as starting LUTs as future upgrades since each

of these LUTs can be

xed GCR =

fine tuned intelligently using carefully designed experiments in
the factory. For example, a maximum

= =

minimum black solution can be easily
created by adjusting the parameters of the GCR equations and still achieve full CMYK
gamut utilization for images by maximizing the accuracy for all the in-gamut node
colors. These and other extensions to multiple GCR

medium

= UCR combinations can be

comprehended within the single initial LUT structure.