Differences in color image reproduction depending on the output device: monitor vs. printer.

Kurecic, Maja Strgar ; Agic, Darko ; Mandic, Lidija 等

1. INTRODUCTION

In the reproduction of color images the one of the most important parameter is the gamut of the color reproduction media. Color reproduction media is term used for describing the media for capturing or displaying color information, e.g. a digital camera, scanner or a CRT monitor. In printing, the color reproduction media is not the printer itself, but the combination of printer, colorants and substrate (Morovic & Sun, 2002). According to the terminology used by CIE Technical Committee 8-03 on gamut mapping (http://www.colour.org/ tc8-03), the color gamut is a range of colors achievable on a given color reproduction media under a given set of viewing conditions.

When discussing the color gamut it is very important to specify the corresponding viewing conditions and observer details. For example, printed image has a set of possible color gamuts, rather then a single one. Different levels of illumination result in different gamut volumes, illumination chromaticity changes gamut shape as well as volume. Viewing distance and flare in the environment also affect the color gamut (Morovic, 2003). Another thing to consider is that some output devices are direct color reproduction media (like the monitors), while others (like printers) are not. As mentioned earlier, printers are not themselves color reproduction media, so it is not possible to talk about printer's gamut, if details about substrates, colorant, viewing conditions and observer are not specified. And, even then, it is more appropriate to talk about the color gamuts of prints, not printer.

Practical implications of above given information's are when perform device characterization in the ICC color managed workflow. It is essential that each media is characterized under the individual combination of observer and viewing conditions for which the resultant device profile will be used. For example, characterizing a CRT monitor in a dark, flare-free environment, and then using it under typical office conditions is invalid and will result in substantial error.

Among the variety of devices, the gamut difference is most noticeable between a CRT monitor and a hardcopy printer. For a typical printer, the gamut volume in CIELAB color space is only 50 to 80 % of that of a typical CRT monitor (Katoh, 1997). A monitor's gamut is wider particularly in the green and blue regions and in the high lightness region. For this reason, it is impossible for the printer to reproduce all the colors achievable on the monitor. Consequently, out-of-gamut colors of the monitor have to be mapped to the inside of the printer gamut while minimizing a change in image appearance. The purpose of gamut mapping is to preserve the appearance of an image as much as possible when the image is reproduced by a different device or in a different media (Fraser et al., 2005). Beside output devices, a visual characteristic of image that is being reproduced also influences how an image will be reproduced by different gamut mapping algorithms.

2. EXPERIMENT

In this experiment, the images (originals) intended for reproduction process were four gouache paintings. Chosen paintings have very different visual characteristics; their dominant colors, lightness range and saturation of used paints were different. Intention was to investigate the influence of specific visual characteristics of the original on the color reproduction. The color target used for digital camera characterization and the paintings were photographed in controlled studio conditions. Raw files (12 bits) from camera was converted in TIFF format and then supplied to the Profilemaker software, which relates camera RGB signals to measured CIELAB values from target. Digital camera profile was made for light source used for imaging (daylight). The output device used was cathode ray tube (CRT) monitor. Device was calibrated and characterized with GretagMacbeth Measure Tool and ProfileMaker software in darkened room. Second output device used in this experiment was Epson Stylus PRO 4000 printer. Device was calibrated and characterized in the way to enable the largest gamut possible with given combination of paper and ink. Files coming to output device were in RGB mode, and were converted in CMYK with EFI Color Proof XF RIP. All reproductions used in evaluation were printed in the same day. Color reproduction media with specifications used in the experiment are listed in Tab. 1.

3. RESULTS AND DISCUSSION

On the basis of spectrophotometric measurements taken from monitor and from print, device gamuts were calculated and expressed in Color Cubic Units (CCU) for CIELAB color space, using ColorThink Pro application. The volume comparison of color gamut of used input and output devices is shown in Tab.2.

To be able to quantitatively measure the color accuracy of reproductions, it was necessary to generate evaluation target for every painting, by choosing 24 different colored areas from digital image of painting. Those areas were measured with GretagMacbeth Spectrolino spectrophotometer directly from original painting using mask with holes at exact coordinates, and compared with values of color samples (on evaluation target) measured from reproductions on monitor and reproductions on prints. Color difference ([DELTA][E.sup.*.sub.94]) and differences in lightness ([DELTA][L.sup.*]), chroma ([DELTA][C.sup.*.sub.ab]) and tone ([DELTA][H.sup.*.sub.ab]) for every original and reproduction were calculated and compared, shown in Tab.3. A larger [DELTA][E.sub.94] is indication that there is a more color variation. A negative [DELTA]L and [DELTA]C values indicate the reduction of lightness and chroma on reproduction.

To compare the colors of the original and colors of the reproductions, distribution of color samples from Painting 3 was plotted in CIELAB diagram, shown in Fig. 1. Paintings reproduced on monitor shows significantly smaller reduction of lightness and chroma then those reproduced on prints. Large errors in chroma are present mostly for orange, dark blue and green colors, which are out of printer gamut. Differences in tone are also larger for reproductions on prints. Overall best reproductions are those of Painting 1, which have low saturation and narrow lightness range, with uniformity of tone values.

Total color differences obtained for samples reproduced on monitor are smaller than those obtained for reproductions made by printer. The reason for that is the fact that CRT monitor's gamut is larger than printer's gamut (Tab.2.), so as the possibility of color accurate reproduction, especially for very saturated colors. It is important to emphasize that those color differences obtained for reproductions made by printer would be even larger if the printer were used as hard proof for offset press. In that case, different device profile would be used (e.g. EuroscaleCoated.icc) with significantly smaller color gamut volume (375,318 CIELAB CCU).

[FIGURE 1 OMITTED]

4. CONCLUSION

Observed color differences in image reproduction on CRT monitor and on print arise from the fundamental dissimilarities of those media. Monitor is self-luminance device, which produces color by mixing colored lights, using additive color mixing, while printer produce color by mixing of colorants, using subtractive color system on reflective substrate. Emissive devices like monitors attain their highest chroma at a relatively high lightness compared to the reflective media like prints. So, it is impossible to match output colors (on monitor or print) with original image colors exactly. There will always be some colors that are not reproducible by any output device. The quality of their reproduction will mostly depend on improved gamut mapping algorithms, which will be the subject of our future research. The next step will be to prove it with various reproduction systems.

5. REFERENCES

Fraser, B.; Murphy, C. & Bunting, F. (2005). Real World Color Management, Peachpit Press, ISBN 0-321-26722-2, USA

Katoh, N. (1997). Practical method for appearance match between soft copy and hard copy, Proceedings of SPIE, Walowit, E. (Ed.), pp. 170-181, SPIE Publication Vol. 2170, ISBN 9780819414656, USA

Morovic, J & Sun, P. (2002). How different are colour gamuts in cross-media colour reproduction; Chapter 12 in Colour Image Science, edited by MacDonald, L. W. & Luo, M. R., Jon Wiley & Sons Ltd, ISBN 0-471-49927-7, England

Morovic, J. (2003). Gamut mapping; Chapter 10 in Digital Color Imaging Handbook, edited by Sharma, G., CRC Press LLC, ISBN 0-8493-0900-X, USA

*** (2004) http://www.colour.org/tc8-03, Gamut Mapping, Accessed on: 2008-11-19

Tab.1. Color reproduction media with specifications used in
experiment

Color reproduction Commercial name Specifications
media

Input device Nikon D70 digital Used studio
 camera with AF illumination:
 Zoom-Nikkor 35-70mm Dedolight DLH 200S,
 objective daylight 5500K, 245
 lux, 3m distance
 from object

Output device Mitsubishi Diamond White point: D65,
 PRO 2070[SB] CRT Gamma: 1.8,
 monitor Brightness: 100%

Output device Epson Stylus PRO Pigment toner on
 4000 printer EFI Gravure Proof
 Paper 4245 semimatt,
 245 g/[m.sup.2]

Tab. 2. Volume comparison of color gamuts of digital camera,
monitor and print

 Device profile Gamut volume

Digital camera CGRT24D.icc 1.006.260
Monitor ICC monitor 0802.icc 843.734
Printer EPSON 4000.icc 667.660

Tab. 3. Comparison of mean values for differences in lightness
([DELTA][L.sup.*]), chroma ([DELTA][C.sup.*.sub.ab]), tone
([DELTA][H.sup.*.sub.ab]) and color difference
([DELTA][E.sup.*.sub.94]) for color reproductions on monitor and print

 ORIGINAL [DELTA] [DELTA]
 [L.sup.*] [C.sup.*.sub.ab]
 mean mean

MONITOR Painting 1 -0.12 1.07
 Painting 2 0.04 -2.02
 Painting 3 -1.94 -4.92
 Painting 4 -0.62 -2.80
 Final result -0.66 -2.17

PRINTER Painting 1 -0.58 -2.30
 Painting 2 -1.72 -3.23
 Painting 3 -4.59 -9.82
 Painting 4 -1.71 -6.75
 Final result -2.15 -5.53

 ORIGINAL [DELTA] [DELTA]
 [H.sup.*.sub.ab] [E.sup.*.sub.ab]
 mean mean

MONITOR Painting 1 2.21 3.23
 Painting 2 2.90 4.71
 Painting 3 2.94 4.53
 Painting 4 2.41 3.24
 Final result 2.62 3.93

PRINTER Painting 1 2.71 3.99
 Painting 2 4.49 6.79
 Painting 3 3.86 7.65
 Painting 4 3.37 5.88
 Final result 3.61 6.08