Using measured photography to obtain optimal results from CCD Color Scanners

Show full item record

Title: Using measured photography to obtain optimal results from CCD Color Scanners
Author: Milburn, David L.
Abstract: With the advent of desktop color scanners in the electronic prepress production environment, quality in color separations from transmissive originals has become erratic. The cause of this fluctuation is, in part, due to the transition from PMT to CCD-based scanning technology. Whereas, PMT scanners tend to have a broad dynamic range, that of CCD scanners is more limited. This characteristic adversely affects the quality of color separations by causing additional tone compression. An original transparency typically has a shadow density of 3.00 and a diffused highlight density of 0.30 for an overall density range of 2.70. On a four-color heatset web press with coated stock, the maximum reproducible tonal range corresponds to a density of 1.80. The difference in density of 0.90 between the original and the press sheet is unable to fit through the printing window unless it undergoes considerable tone compression. This project was based on two hypotheses. The first was that the lower the tonal range of a transmissive original the more readily lower-midtone-to-shadow tonality could be retained in the separations produced by a CCD scanner and related equipment. The second was that exposure latitude in the separations would decrease with increasing tonal range. The first stage of production was to produce twelve test transparencies by photograph ing a still life set to four tonal ranges: 3.5, 4.5, 5.5, and 6.5 f/stops. Within each range, three images were selected to represent normal exposure, 1/2 f/stop overexposure, and 1/2 f/stop underexposure. Comparison of halftone proofs, made from separations of the normallyexposed transparencies, were later used for the first hypothesis. Proofs from the 1/2 f/stop over- and 1/2 f/stop underexposures were compared with the normal exposures to test the second hypothesis. vm The twelve test transparencies were first scanned on the Dainippon Screen SG-608 to produce a set of best-of-kind reference separations and halftone proofs. Next, the trans parencies were scanned on two midrange 12-bit CCD scanners, one a Pixelcraft CIS 4520RS, the other an Agfa Horizon. Separations for both were produced with Color Access 1.3.3 software on a Macintosh Quadra 700 computer, linked to an Agfa SelectSet 5000 image-setter via an Agfa 5000PS Star Plus RIP. The image files were placed in an 8 1/2" X 11" QuarkXPress page with a 20% gray surround prior to output. Halftone proofs were produced with the Fuji Color Art proofing system, then viewed under 5000 Kelvin lighting. Three methods were used for comparison: visual evaluation by the author, densitometric measurement, and evaluation by 32 independent judges. Only proofs from the two CCD scanners were shown to the judges. Proofs from the SG-608, of noticeably higher quality, were used for reference. The four proofs from separations produced by normally-exposed originals were used to examine the first hypothesis. Two groups of proofs, one for each scanner, were ranked by the judges according to best-to-worst rendition of lower-midtone-to-shadow detail. The rank ings for both groups placed the 3.5 f/stop tonal ranges first, 4.5 f/stop second, 5.5 f/stop third, and 6.5 f/stop fourth. Visual evaluation by the author ranked the proofs in the same order, establishing a 100% correlation. Increases in density range were also expected to follow the 3.5 to 6.5 f/stop ranking. But actual measurements showed increases in density to 4.5 f/stops, then a pronounced drop of 0.25 or greater for the 5.5 and 6.5 f/stop ranges. This demonstated that the higher tonal ranges exceeded the capacity of the CCDs to make a full response, indicating that limited dynamic range was causing abrupt increases in tone compression. To determine the validity of the second hypothesis, the judges examined two groups of twelve proofs each, corresponding to the separations for normal, 1/2 f/stop over- and 1/2 f/stop underexposed originals within each of the four tonal ranges. For the Agfa Horizon, the rankings were 3.5 f/stop tonal range first, 4.5 f/stop second, 5.5 f/stop third, and 6.5 f/stop fourth. With the Pixelcraft CIS 4520RS, the rankings were 3.5 f/stop first, 6.5 f/stop second, 5.5 f/stop third, and 6.5 f/stop fourth. The misranking of the 6.5 f/stop range in second place indicated the difficulty the judges experienced in distinguishing between higher tonal ranges due to the increasing effects of tone compression. Again, densitometric measurements did not support the rank ings of the judges or the author because the densitometer could not distinguish between small tonal range differences due to good exposure latitude and those differences due to blocked shadow tonality resulting from tone compression. A procedure was devised for mathematical assessment of tonal range differences. Using the change in density between adjacent tonal ranges, a value was derived which could be expressed as a fraction of the inital 1 f/stop difference between ranges ( 1 f/stop = 0.30 den sity units). For example, the 3.5 f/stop normal exposure for the Agfa Horizon had a tonal range of 1.73 and the 4.5 f/stop normal exposure had a tonal range of 2.02. The change from 3.5 to 4.5 f/stops is +0.29. Expressed as a percentage, 0.29 -0.30 = 97%. This value showed that the scanner made a nearly complete response in translating tonal range differences in the original transparency to tonal range differences in the separations. Further refinement of these calculations is needed to reflect differences in gamma and tone compression between the original transparencies and the halftone proofs. The findings of this study demonstrate that optimal results from a CCD or PMT scan ner can only be obtained if the tonal range of the original scene is less than 4.5 f/stops. When the 4.5 f/stop range was exceeded, the CCD scanners responded with an immediate deterioration of lower-midtone-to-shadow tonality. In comparison, PMT scanner displayed a more gradual degradation of lower-midtone-to-shadow tonality, in keeping with its greater dynamic range.
Record URI: http://hdl.handle.net/1850/11095
Date: 1993-11

Files in this item

Files Size Format View
DMilburnThesis11-1993.pdf 19.55Mb PDF View/Open

The following license files are associated with this item:

This item appears in the following Collection(s)

Show full item record

Search RIT DML


Advanced Search

Browse