EE 638: Principles of Digital Color Imaging Systems Module 32

Recall: Primaries Additive systems (display devices) Ideal block model for primaries Subtractive color model - Reflectance

Consider ideal block subtractive primaries Let Consider ideal block subtractive primaries Assume spatially non-overlapping primaries i.e. at any point on paper have either Cyan, Magenta, Yellow, or no colorant; can write for locally average absorptance: C=B+G M=B+R Y=G+R c, m, y – fractional area coverage of each colorant

Printing technologies Problem: Real colorants have non-ideal overlapping absorptances Are spatially non-overlapping colorant realistic? Printing technologies Traditional analog printing technologies  plate-based technologies (1) letter press (oldest) http://vandercookpress.info/vanderblog/timeline/ The information about the VanderCook press was provided by Steve Brune, HP Inc., Boise, ID (6/26/17)

Where is letterpress today? text https://www.honors.purdue.edu/community/news/news-archive/honors-college-and-cla-launch-print-labratory/index.php

(2) gravure (3) intaglio ink is contained in wells ink is goes here tone is controlled by locally varying density on ink Used for National Geographic (3) intaglio printed under great pressure Used for currency ink is contained in wells ink is goes here plate

(4) offset lithography - ink - water Printing plate Most common high speed/high volume printing technology - ink - water Printing plate Blanket transfer roller improves quality of transfer paper

Traditional halftone printing Rotate screens with respect to each other by 15º Recall M-D equation for one colorant C Typical halftone print (2 color) Cyan Magenta Average reflectance

text From work of Alty Jumabayeva

text From work of Alty Jumabayeva

Let be total fractional area coverage of Cyan Neugebauer primaries Let be total fractional area coverage of Cyan Let be total fractional area coverage of Magenta Unknown and depends on half-toning algorithm known

Want an expression for in terms of In general, this will depend on the color halftoning algorithm. Assume uniform, independent placement of colorant dots Pick an arbitrary point in square and compute probability that each of four different Neugebauer occurs there: { Cyan is printed at & magenta is not printed at }

Assume independent, random placement of colorants Assume independent, random placement of colorants. For example, for a_c the probability that a pixel is covered by cyan only is probability that it is covered by cyan times the probability that it is not covered by magenta. Note that this does not reflect the structure of any particular halftoning algorithm. De Michel coefficient

Assume independent, random placement of colorants Assume independent, random placement of colorants. For example, for a_c the probability that a pixel is covered by cyan only is probability that it is covered by cyan times the probability that it is not covered by magenta. Note that this does not reflect the structure of any particular halftoning algorithm. De Michel coefficient

is spectral reflectance of a solid patch of cyan  Next step: Find CIE XYZ coordinates corresponding to our printed patch with colorant amounts C & M and spectral reflectance Assume illuminant power spectral density is spectral reflectance of a solid patch of cyan 

For example, are the Tristimulus coordinates for solid cyan printed on a particular media and viewed under illuminant I. What do we have? Have model for Note: are measurements of appropriate colorant combinations made when they are printed on a particular media (substrate) (Forward printer model) C, M, Y, K Printer XYZ by extension

What about dot gain? Mechanical dot gain adjust, C, M, Y, K or more directly adjust etc. Optical dot gain apply Yule-Nielsen model: N – empirically determined parameter of the model

Extension to 3-color printing Neugebauer primaries C, M, Y CM, CY, MY CMY, W B G R Kc For four color printing, just extend model straight forwardly by adding a fourth primary Kp process black Why use 4-color system? Kp is cheaper than Kc Minimizes mis-registration impact (important for text) Bigger gamut (specifically darker or graphics colors can be printed. CMYK yields rich black) || || || || primaries secondary tertiary pros

cons No longer have unique 1-to-1 mapping from printer-colorant-in to CIE-XYZ-out More expensive cons

Y M C W Neugebauer model Assume random placement of colorant De Michel coefficient