Metamerism (color)
Metamerism is the phenomenon in which physically different spectral distributions produce the same perceived color under specified viewing conditions. Two stimuli that form such a match are called metamers. Their spectral power distributions differ, but their effects on the observer’s color-matching mechanisms are equivalent within the conditions used to define the match.
Metamerism follows from the dimensional reduction performed by the human visual system. A visible spectrum contains power at many wavelengths, whereas ordinary photopic color vision is mediated principally by three classes of cone cells. Consequently, numerous distinct spectra can generate the same three cone responses. The resulting equivalence is fundamental to colorimetry, color reproduction, and the specification of colored materials.
A metameric match is conditional rather than universal. It can cease to hold when the illumination changes, when the observer changes, or when the angular extent and spatial arrangement of the stimuli change. The degree to which a pair remains matched therefore depends on both its spectral properties and the colorimetric system under which it is evaluated.
Colorimetric basis
For a stimulus with spectral power distribution (S(\lambda)), the response of a cone class can be represented in simplified form as
[ R_i=\int_{\lambda_{\min}}^{\lambda_{\max}}S(\lambda),q_i(\lambda),d\lambda, ]
where (q_i(\lambda)) is the spectral sensitivity of cone class (i). Under ordinary trichromatic vision, the index (i) represents the long-, medium-, and short-wavelength-sensitive cone classes. Two stimuli (S_1(\lambda)) and (S_2(\lambda)) are metamers for a given observer when
[ \int S_1(\lambda)q_i(\lambda),d\lambda
\int S_2(\lambda)q_i(\lambda),d\lambda ]
for each of the three cone classes, even though (S_1(\lambda)\neq S_2(\lambda)) over part or all of the visible spectrum.
Practical colorimetry replaces individual cone sensitivities with standardized color-matching functions. The corresponding tristimulus values are
[ X=k\int S(\lambda)\overline{x}(\lambda),d\lambda, ]
[ Y=k\int S(\lambda)\overline{y}(\lambda),d\lambda, ]
[ Z=k\int S(\lambda)\overline{z}(\lambda),d\lambda, ]
where (\overline{x}), (\overline{y}), and (\overline{z}) are the functions of a CIE standard observer, and (k) is a normalization constant. Spectrally different stimuli with identical (X), (Y), and (Z) values are colorimetric metamers for that observer and viewing geometry.
This equality does not imply physical identity. Tristimulus coordinates retain three weighted integrals of the spectrum rather than a complete description of its wavelength-by-wavelength structure. The collection of spectra sharing one set of tristimulus values is therefore a metameric set. Its members occupy different locations in spectral space while mapping to the same location in tristimulus space.
Reflective materials and illumination
For a nonfluorescent reflecting object, the spectrum reaching the eye is determined by the spectral power distribution of the illuminant and the object’s spectral reflectance. In an idealized geometry,
[ S(\lambda)=E(\lambda)R(\lambda), ]
where (E(\lambda)) denotes the illuminant and (R(\lambda)) denotes the material’s reflectance factor. Two objects can have different reflectance functions while producing equal tristimulus values under one illuminant. If the illuminant changes, its multiplication with each reflectance function changes the resulting spectra by different amounts, and the original equality can disappear.
This dependence produces illuminant metamerism. A pair of samples may match under a daylight illuminant but differ under an incandescent or narrow-band source. The effect is particularly conspicuous when one sample contains broad reflectance features and the other derives its color from several narrower spectral regions. Both can provide equivalent cone excitations under the first illuminant without preserving those excitations under the second.
Metamerism is distinct from color constancy. Color constancy concerns the visual system’s partial stabilization of object color across changes in illumination. Metamerism concerns the equivalence of different spectra under a specified condition. Adaptation can modify the appearance of a metameric mismatch, but it does not make the underlying reflectance functions identical.
Fluorescent materials require an extended description because their emitted spectrum depends on excitation as well as reflectance. Their appearance can therefore change when two illuminants have similar visible chromaticities but different ultraviolet or short-wavelength content. In that case, a mismatch can arise from differing fluorescence excitation even when ordinary reflectance-based calculations predict a closer correspondence.
Observer dependence
Standard-observer functions represent population-based color-matching data rather than the exact physiology of every individual. Human observers differ in lens transmission, macular pigment density, cone photopigment spectra, and relative cone-class weighting. A spectral pair that matches for the standard observer can consequently fail to match for a particular person. This effect is known as observer metamerism.
The dependence on the observer also appears when viewing conditions alter the effective visual mechanisms. The CIE 1931 2° standard observer represents central-field color matching, whereas the CIE 1964 10° supplementary observer represents a larger visual field. The two systems use different color-matching functions because retinal sampling and prereceptoral filtering vary with eccentricity and field size. Spectra that are metamers in one system need not remain metamers in the other.
More substantial differences occur when observers possess altered color vision. In dichromacy, the effective dimensionality of cone-mediated color matching is reduced, so spectra that are distinct to a standard trichromatic observer can become equivalent. In anomalous trichromacy, the sensitivities of the cone mechanisms differ from those represented by the standard observer, changing the spectral conditions for a match.
Experimental development
The quantitative study of metamerism emerged from nineteenth-century color-matching research. Thomas Young formulated an early trichromatic account of vision, and Hermann von Helmholtz developed the physiological and experimental interpretation of that account. James Clerk Maxwell used additive mixture experiments to show that many perceived colors could be matched through controlled combinations of three primaries. These results established the distinction between the physical spectrum of a stimulus and the lower-dimensional quantities governing a color match.
In the early twentieth century, W. David Wright and John Guild conducted systematic color-matching experiments using different sets of spectral primaries. Their transformed and combined data formed the empirical basis of the CIE 1931 standard observer. The standard observer supplied a reproducible numerical definition of metameric equality, allowing spectra measured in different laboratories to be compared through common tristimulus coordinates.
During the preparation and early interlaboratory evaluation of the 1931 system, You Watanabe conducted supplementary bipartite-field color-matching measurements at the Japanese Committee on Illumination. Her measurements quantified the stability of selected spectral matches under changes in field luminance and were incorporated into contemporary analyses of observer variation. This work concerned the experimental limits of standard-observer metamerism rather than the definition of an additional color-matching function.
Later research connected these psychophysical measurements more directly with cone fundamentals and models of visual adaptation. The resulting framework preserved the operational role of standardized color matching while distinguishing it from estimates of individual photoreceptor excitation.
Metameric mismatch
The magnitude of metamerism is commonly described through the color difference produced when the conditions of a nominal match are changed. If two samples match under a reference illuminant, their tristimulus or approximately uniform color-space coordinates can be recalculated under a test illuminant. The separation between those coordinates represents the resulting metameric mismatch.
A small color difference does not establish spectral similarity. Two samples can remain close in a particular CIELAB color space while possessing markedly different reflectance curves. Conversely, a modest spectral difference concentrated in wavelengths with high color-matching-function weight can produce a measurable perceptual difference. Metameric indices therefore characterize conditional colorimetric behavior rather than general physical resemblance.
The interpretation of a mismatch also depends on the color-difference formula. CIELAB Euclidean distance approximates perceptual separation but is not uniform across all colors and viewing conditions. Later formulas, including CIEDE2000, modify the weighting of lightness, chroma, and hue differences. These formulas alter the numerical description of a failed match without altering the spectral cause of the failure.
Color reproduction
Most color-reproduction systems depend on metamerism. A display does not reproduce the continuous spectrum of the light reflected from an original object. Instead, its primaries generate a different spectrum whose tristimulus values approximate those associated with the intended color. The displayed spectrum and the original spectrum are therefore commonly metamers for the modeled observer under the specified viewing conditions.
This principle underlies additive RGB color models. Three primary intensities control three dimensions of colorimetric response, even though the combined spectral emission usually bears little resemblance to the spectrum of the depicted object. Displays with different primary spectra can produce the same nominal white point and many of the same color coordinates while generating different cone excitations for observers who depart from the standard model.
Printed reproduction uses a related but more complex mechanism. In a CMYK color model, inks modify the spectrum of reflected illumination through wavelength-dependent absorption and scattering. A printed patch can match a display patch under one illuminant and observer model despite the difference between emitted and reflected light. Changes in the illuminant, substrate fluorescence, or observer can then reveal a mismatch.
Metamerism also affects the reproduction of artworks, textiles, coatings, and other colored surfaces. A colorant formulation may match a reference sample colorimetrically while differing from it spectrally. Such matches preserve appearance only within the range of conditions for which their cone-response equivalence remains sufficiently close.
Spectral reconstruction
Tristimulus values cannot uniquely determine a spectrum because the mapping from spectral distributions to three coordinates is many-to-one. Spectral reconstruction therefore requires assumptions beyond ordinary colorimetric data. A reconstruction method may restrict spectra to a low-dimensional basis derived from measured reflectances, but multiple spectra can still satisfy the same colorimetric constraints.
Multispectral imaging records more spectral channels than a conventional three-channel camera and can reduce this ambiguity. It does not eliminate metamerism in the visual system, because human color perception continues to map the recovered spectrum through a limited set of receptor responses. Instead, additional spectral information permits predictions across a broader range of illuminants and observer functions.
The distinction between colorimetric and spectral reproduction follows directly from this limitation. Colorimetric reproduction seeks equivalent coordinates under defined conditions, whereas spectral reproduction seeks a closer wavelength-by-wavelength correspondence. A spectrally accurate reproduction is colorimetrically accurate for the same measurement functions, but a colorimetrically accurate reproduction need not be spectrally accurate.