Color Reproduction
Color Reproduction
Paintings reproducing scenes date back 160 centuries to colored wall art in the Lascaux caves. Printing has early examples of stone characters for making clay tablets dating back 50 centuries; color wood block plates for printing uniform colors 14 centuries; Gutenberg’s press nearly 5 centuries; and LeBlon’s hand-made color-separation printing plates 3 centuries -- 160 years before James Clerk Maxwell invented the first mechanical color photography.
All human color appearances can be described in a three-dimensional color space, with amounts of red, green and blue light as axes. Replicas are copies of an original using the same medium, or color materials, having the same physical properties for controlling light. Replicas share the exact same color space. Using the same materials means the color reproduction can be exact.
Reproductions are copies of original art in a different medium, such as a computer-screen copy of a painting. The challenge in color reproduction is to copy the information contained in the entire original 3-D color space into a different size and shape reproduction space. The problem is very similar to moving. The original is one’s current house. (It is defined by the all the tools of physics used to measure light, such as wavelength, amount of light, etc.). The “reproduction” house has different dimensions for the length (amount of red), width (amount of green) and height (amount of blue). Reproductions move everything in the old house into the new house, keeping all contents in corresponding rooms, even though the dimensions of the entire house, and each room, are different. Good reproductions are never exact physical copies of the original, because that is impossible. Good reproductions capture the appearance and relationships of objects in the scene. The original and the reproduction have different 3-D color spaces with different sizes and local shapes, because they use different colorants. Good reproductions reproduce all the information in the interior of the original’s color space using completely different colorimetric colors at corresponding pixels.
Real scenes have the greatest range of light. Light emitting displays (TVs) have much less range, yet more than reflective prints. Dynamic ranges characterize the light reproduction values between the most and least response for each RGB channel, for each media. Color saturation characterizes the entire range of chroma of the medium’s colorants. Together range and chroma determine the color gamut, that is, all possible light values, of the medium. With greater color saturation of the dyes and filters, we find larger color gamuts. However, these pure colors are just 3, or 4, points in the entire color space. While the dyes and filters control the extreme boundaries, the quality of the reproduction depends on how the media presents the scene information in the interior of the color space. The term “tone scale” is used to describe how the media processing alters the interior of the original’s color space in the reproduction.
Color forming technologies
The simplest and oldest color technology uses opaque colorants. Here, the light is reflected from the front surface. Opaque pigments contain materials that scatter light so as to prevent light from interacting with subsurface colorants.
James Clerk Maxwell’s invention of additive color photography used three black-and-white separation photographic transparencies taken with red, green and blue filters. He projected the three images in superposition using the same filters used to take the pictures. The photographs recorded the information from the scene in each spectral region. By projecting all three records in superposition the reproduction adds the information from each waveband. White is the sum of all the light in all three. Black is the minimum transmission for all three wavebands. All other colors are additive depending on the relative transmissions, in the RGB records.
Subtractive color reproductions, first demonstrated by du Hauron, use the same RGB color separation information as additive. Here colors are controlled by multiple absorptions (transparent dyes on top of other dyes). Each transparent subtractive dye absorbs one-thirds of the visible spectrum, namely, minus-red(cyan), minus-green(magenta), and minus-blue(yellow). Each subtractive dye modulates the R, G, or B information.
(Left), the absence of dye makes white in the print by reflecting R, G, B light. (Right), the cyan dye absorbs red, so that both green and blue light is reflected to the eye.
The figure above illustrates the transmission and absorption of light in a cross-section of a hypothetical three-layer color reflection print. The colors were taken from the horizontal piece in the above subtractive illustration - divided into three magnifications.
(Left), the combination of cyan and yellow makes green.
(Right), the superposition of all three makes black.
(Left), the combination of magenta and yellow makes red.
(Right), magenta alone reflects red and blue light.
Most subtractive printing systems, inkjet and printing, use a fourth print ink, black. This has practical and financial advantages. Practically, it makes achromatic blacks and grays with a single colorant, rather than precisely controlling the amounts of yellow, magenta and cyan. Economically, at least in printing, black inks are less expensive, than colored inks.
Spatial additive color in Flat-Panel Displays
Prior to the development of subtractive Kodachrome three-layer dye systems in 1935, single-exposure color films used additive, side-by-side color forming systems. These systems included Autochrome (dyed starch grains with silver halide salts), Dufy, and Jolly three color patterns. These are the direct ancestor of three gun color television and flat-panel Liquid Crystal Displays (LCD). They use the same spatial additivity to display colors. Here the display has small transparent R, G and B filters. The RGB signals sent to each triplet controls the amounts of R, G, and B light. This process is called additive reproduction because the side-by-side triplets of R, G, and B stripes are too small (100 triplets/in) for the observer to resolve at normal viewing distance. These stripes get blurred by the eye to make a spatially additive reproductions. Look at an LCD display, or LCD television with a 10x-magnifying lens to see the additive substructure.
Tone scale control of the interior color space
So far, we have discussed how to form three, or four, colors on the boundary of the color gamut. All the other colors are formed by the controlled combinations of these few colors. Each technology uses different tone-scale manipulations to manage the interior of the color space.
Color Photography
Color photography uses three transparent dyes on transparent or reflective support sheets. Light activates the silver halide sensors that in turn control the amount of dye at each pixel. There is very little micro-structure near visual resolution threshold.
Printing
Halftone color printing is the most commonly used technique in both computer printers and commercial publications. It reduces the amount of light coming to the eye by increasing the area of light-absorbing ink. White is no dot; black is 100% dot. Most colored halftone uses yellow, magenta, cyan and black dots.
Optimizing color halftone images involves many factors. They include: the colorants’ spectra, the colorants’ covering power, the colorants’ opacity, the paper’s spectra, the paper’s surface, the presence of optical brighteners (fluorescent dyes in the paper), the size and shape of the dot patterns, the dot placement relative to each other, and dot overlap. Hans Neugebaurer developed a model to calculate color response of printing systems. By measuring the response of a subset of possible printed colors, one can combine the interactions of all these complex variables. These measurements include: no ink, 50%, and 100% of all inks separately, and in all combinations. In other words, by printing selected samples of all-possible combinations, we can accurately calculate the complex system responses of its 3-, or 4-dimensional color space.
Colorimetric Reproductions
Using a different medium usually makes it impossible to accurately substitute colorimetric matches for every picture element (pixel). If the reproduction color space is different, (the room is different), then exact color reproduction techniques create errors that are easy to see, because parts of the original’s information are left out. (All the furniture does not fit in the different room.)
One might think that the answer to successful reproduction is to just rescale all values between maxima and minima to fit the new dimensions. If the reproduction space is smaller that means a lower-contrast copy. However, observers prefer enhanced reproductions to accurate renditions, and dislike low-contrast renditions. Near white, near black and along the high-chroma color gamut, reproductions change more slowly than do originals (lower contrast). In the middle of the color space preferred reproductions have higher contrast. These non-linear tone-scale curves help to preserve spatial information and reduce artifacts. Reproductions capture spatial color relationships of the original and render their copies as approximate visual appearances. High-quality reproduction requires both high-chroma colorants for color saturation, and carefully crafted tone-scale transformations within the color space. The difference between a good reproduction and a bad one is how well the tone-scale alters the original to synthesize combinations of RGB colorants that appear the same as the scene using different stimuli.
Reproduction of Scene Dynamic Range
Some reproductions involve small transformations of the interior of the original’s, color space, such as a photographic print of an oil painting in uniform illumination. Real-life scenes, both indoor and outdoor, are almost always in non-uniform illumination. Illumination introduces a major challenge to reproduction by having an extremely large original color space. On a clear day shadows cast by the sun are 30 times darker than direct sunlight. Real-life scene reproduction is analogous to moving a castle into a cottage.
Imaging techniques can record scene information over a High Dynamic Range (HDR) of light. The range of captured information is much more than the 30:1 range possible between white and black in a reflective print. Ansel Adam’s Zone System provides the logical framework for capturing the wide range of light in natural scenes and rendering them in a smaller dynamic-range print. Adams described a three step process: measuring scene range, adjusting image capture to record the entire scene range, and locally manipulating the print exposure to render the high-range scene into the low-range print. Adams visualized the final image before exposing the negative. He assigned appearances from white to black to image segments. Once the negative recorded all the information, he controlled the local print contrast for each local part of the image (manually dodging and burning) to render all the desired information from a high dynamic range scene into a low-dynamic range print. Not only can these techniques preserve detail in high- and low-exposures, they can be used to assign a desired tone value to any scene element. Adams described the local contrast control in detail for many of his most famous images. Adams developed chemical and exposure manipulations to spatial control appearances between white and black. Today, we use Photoshop in digital imaging.
Painters have used spatial techniques since the Renaissance to render HDR scenes, and photographers have done so for 160 years. Land and McCann’s Retinex algorithm used the initial stage of Adam’s wide-range-information capture for its first stage. Instead of using aesthetic rendering, it adopted the goal that image processing should mimic human visual processing. The Retinex process writes calculated visual sensations onto prints, rather than writing a record of light from the scene. To this aim, Retinex substitutes the original light values at each pixel with ratios of scene information. This approach preserves the content in the original found in the interior of the color space.
Electronic imaging made it possible, and practical, to manipulate images spatially. Automatic spatial processing is not possible in silver halide photography because film responds locally. Silver grains count the photons. The same quanta catch produces the same film density. Hence, Adams had to manipulate his images by hand. Digital image processing, or its equivalent, had to be developed in order for each pixel to be able to influence each other pixel, as in human vision. Details in the shadows are necessary to render objects in shade to humans. The accuracy of their light reproduction is unimportant: the spatial detail of objects in shadows is essential. Spatial-comparison image processing has been shown to generate successful rendering of HDR scenes. Such processes make use of the improved differentiation of the scene information. By preserving the original scene’s edge information, observers can see details in the shadows that are lost in conventional imaging.
A good color reproduction conveys the appearance of the original. The interesting feature of excellent copies is that they do not reproduce the original’s stimulus. They cannot, any more than the mover can exactly reproduce the original house in a different size reproduction house. The image reproduction industry has developed high-chroma colorants, ingenious tone-scale and spatial-image processing techniques, so that they can make excellent reproductions. The secret is that they do not reproduce the original’s stimulus. What they do is more like rescaling the furniture for each room so that it has the same spatial relationship with other objects in the new size of room. Good reproductions retain the original’s color spatial relationships in the interior of the color space. They do this because they are indifferent to reproducing the light coming from the original. By definition, this is impossible for a reproduction. It is only possible in a replica in uniform illumination.
Magnifications of Halftones
Suggested further readings:
Adams, A. (1961). The Negative. Boston: New York Graphical Society, Little, Brown & Company.
Adams, A. (1984). Examples: The Making of 40 Photographs, Boston: New York Graphical Society, Little, Brown & Company.
Hornak, J. P. (Ed.), (2002). Encyclopedia of Imaging Science and Technology. New York: John Wiley & Sons, Inc.
McCann, J. J, (2005). Rendering high-dynamic range images: Algorithms that mimic human vision. Proc. AMOS Technical Conference, 19–28.
McCann, J. J. (2007). Art, science, and appearance in HDR images. J Soc Info Display 15/9, 709-719.
McCann, J. J. and Miyake, Y. (2008). The Interaction of Art, Technology and Customers in Picture Making, IEICE Trans. Fundamentals, E91-A, N0 6 1369-1382.
Yule, J. A. C. (1967). Principles of Color Reproduction. New York: John Wiley & Sons, Inc.
3D Color Space
An image of a one-inch high newsprint photograph.
A magnification of the helmut.
A magnification of the face guard. The halftone process reproduces colors by the size and placement of Cyan, Magenta, Yellow and blacK (CMYK) dots. For lightest values they are added spatially; for dark values they subtract color.
High Dynamic Range imaging
A. Adams, The Negative, New York Graphical Society, Little, Brown & Company, Boston, 47-97, 1981.