8 vertex HANS: An ultra-simple printer color architecture

The underlying non-linearity of how print colorants combine makes color control in printing significantly more complex than for other color imaging devices. While in additive systems a measurement of their few primaries and per-channel nonlinearities versus luminance is a sufficient basis for predicting color output, printing typically requires the measurement of a large number of colorant combinations. This requirement for many measurements makes accurate color output more challenging and means that setting up a printing system’s color control can be time consuming and costly. The solution presented in this paper involves a new use of the HANS approach, which instead of print optimization looks for simplifying print color formation and therefore also control. In a nutshell this can be achieved by only ever combining eight basic colorant patterns, which results in a display-like color gamut and allows for color control on the basis of their eight measurements and those of the printing system’s optical dot gain. Introduction An important part of the preparing a printer or press for production is to ensure color accuracy and consistency, which are driven by having an accurate ICC color profile (or proprietary device-link) and up-to-date color calibration data respectively. The two aspects of accuracy and consistency are typically split since the former tends to require hundreds to thousands of color patches being printed and measured while the latter can be achieved using tens to hundreds of samples, depending on the specific marking engine and printing system design. Since profiling and calibration are costly, their frequency has to be balanced against the customer complaints that they can help to avoid. Profile and calibrate too rarely and you risk more jobs that have to be re-printed with manual intervention, do it too often and you eat unnecessarily into your profits. The underlying challenge here is that print color control requires large numbers of color patches to be printed and measured, to obtain a good characterization of what colors will result for different digital inputs. The root cause of these requirements is that printing systems are ultimately controlled in terms of colorant amounts, which are very non-linearly and multi-dimensionally related to color (Yule and Nielsen, 1951). Furthermore, these colorant amounts are the result of a color separation mapping values from a device color space like CMYK or RGB, on top of which ICC profiles are built. Such device color spaces are even further removed from colorant patterns on media, which makes the relationship between them and print color even more complex. As a consequence, large numbers of samples are needed to accurately represent the multi-dimensional non-linearity between either colorant or device color space vectors and print color and while there have been several attempts to optimize these (e.g., Monga and Bala, 2008; Morovič et al., 2010; Bianco and Schettini, 2012), the end result is still in the high tens or low hundreds. Instead of taking a printing system as is and looking for ways to optimize its color characterization and calibration, the approach taken here follows a fundamentally different path. It asks what printing system could be fully characterized in ways similar to those used for displays and arrives at a particular solution. The following sections will therefore first present a brief summary of Halftone Area Neugebauer Separation (HANS), which enables the new solution, proceed to describing how HANS can be used in a new way to result in a printing system characterized by eight colorant patterns and finally show results for driving a printing system using the new approach. Halftone Area Neugebauer Separation Instead of operating in a colorant space, the key feature of HANS (Morovič et al., 2011) is that it addresses a printing system in terms of Neugebauer Primary relative area coverages. E.g., for a printer with three inks – CMY, it doesn’t specify output only in terms of ink use (e.g., [C, M, Y]=[30%, 40%, 0%]), but instead determines area coverages for the system’s Neugebauer Primaries: e.g., [W, C, M, Y, CM, CY, MY, CMY] = [50%, 20%, 10%, 0%, 20%, 0%, 0%, 0%] (Figure 1). Figure 1. HANS basics. The result is not only greater control and the ability to access a greater variety of halftone alternatives per color (yielding metamers even in the three colorant case (Morovič et al., 2011b)), but also a fundamental additivity. Given Neugebauer Primaries of measured colorimetry and a system with measured dot gain, the color of a particular halftoned colorant pattern is the convex combination of the NPs’ colorimetries in Yule-Nielsen corrected CIE XYZ. Note, that in an actual printing system, its optical and mechanical dot gain are only some of the mechanisms that are involved in color formation, and that the use of a Yule-Nielsen corrected CIE XYZ space is an approximate, whose accuracy for a test system will be reported later in this paper. While all this is on the color formation side, the key to arriving at a more easily characterizable printing system is the fact that with HANS, print is controlled in area coverage terms that do have the above, Color Colorant vector NPac vector Halftone pattern Halftone pattern H A N S C ol or an t