The advent of computers and their impact on the graphic arts and printing industry has, and will continue to, change the methodology of working and workflow in prepress operations. The conversion of analog materials (prints, artwork, transparencies, studio work) into a digital format requires the use of scanners or digital cameras, coupled with the knowledge of output requirements as related to client expectations. The chosen input sampling ratio (sampling rate in relation to halftone screening) impacts output quality, as well as many aspects of prepress workflow efficiency. The ability to predict printed results begins with the correct conversion of originals into digital information and then an appropriate conversion into the output materials for the intended press condition. This conversion of originals into digital information can be broken down into four general components. First, the image must be scanned to the size of the final output. Second, the input sampling ratio must be determined, in relation to the screening requirements of the job. This ratio should be appropriate to the needs of the printing condition for the final press sheet. Third, the highlight, highlight to midtone and shadow placement points must be determined in order to achieve the correct tone reproduction. Fourth, decisions must be made as to the image correction system to be employed in order to obtain consistent digital files from the scanner and prepress workflow. Factors relating to image correction and enhancement include such details as gray balance, color cast correction, dot gain, ink trapping, hue error, unsharp masking, all areas that impact quality. These are generally applied from within software packages that work with the scanner, or from within image manipulation software after the digital conversion is complete. The question of what is the necessary input sampling ratio for traditional AM screening has traditionally been based on the Nyquist Sampling Theorem. The basis for determining input sampling ratio requirements for frequency modulated (FM) screening is less clear. The Nyquist Theorem (originally from electrical engineering and communications research) has been applied to the graphic arts, leading to the general acceptance of a standard 2:1 ratio for most prepress scanning work. The ratio means that the sampling rate should be twice the screen frequency. This thesis set out to determine if there are dif ferences in input sampling ratio scanning requirements, based on the screen frequency rx selection (lOOlpi AM, 1751pi AM and 21|lFM used in this study), when generating films and/or plates for printing, that might question this interpretation of the Nyquist Sam pling Theorem as it relates to the graphic arts. Five images were tonally balanced over three different screening frequencies and six different sampling ratios. A reference image was generated for each condition using the Nyquist Sampling ratio of 2:1. Observers were then asked to rate the images in terms of quality against the standard. Statistical analysis was then applied to the data in order to observe interactions, similarities and differences. A pilot study was first run in order to determine the amount of unsharp masking to use on the images that would be manipulated in the main study. Seven images were pre sented from which four were selected for the final study. Thirty observers were asked for their preference on the amount of sharpening to use. It was found that for this condition (7 images) observers preferred the same amount of sharpening for the 1751pi AM and 21u FM screens, but slightly more sharpening for the lOOlpi AM screen. This information was then applied to the main study images. An additional image previously published was added after the pilot study, as it contained elements not found in the other images The unsharp masking applied to this image was the same as at the time of publication. The main study focused on the interaction of image type, screen frequency and varia tions of input scanner sampling ratios as it relates to output. The results indicated that image type, sampling ratio, sampling ratio - frequency interaction were factors, but fre quency alone was not. However, viewing the interaction chart of frequency and sampling ratio for the 1751pi AM and 21u FM screens alone, an insignificant difference was indi cated (at a 95% confidence level). The conclusion can therefore be drawn that at the higher screen frequencies tested in this study, viewer observations showed that the input sampling ratios should be the same for 1751pi and 21)1 FM screens. Continuous tone orginals should be scanned at a sam pling ratio of 1.75:1. This answered the question of whether FM screening technology can withstand a reduced input sampling ratio and maintain quality, which this study finds cannot. At the lower screen ruling of lOOlpi the input scanner sampling ratio requirement, based on viewer preferences of the five images presented, can be reduced to a 1.5:1.

Library of Congress Subject Headings

Image processing--Digital techniques--Analysis; Color printing; Desktop publishing; Graphic arts

Publication Date


Document Type


Department, Program, or Center

School of Print Media (CIAS)


Sigg, Franz


Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TA1637 .E47 1999


RIT – Main Campus