In 1931, a group of scientists gathered in Cambridge, England, at a meeting of the International Commission on Illumination (CIE). They had spent years analyzing data from color matching experiments conducted by William David Wright and John Guild, who had asked human observers to match monochromatic colors by mixing red, green, and blue lights. The result of that meeting—the CIE 1931 color space—revealed something unsettling: the shape of human color perception is fundamentally incompatible with the triangle-based color systems used by every display today.

Look at a sunset. The deep oranges and reds bleeding into purple. The saturated cyans of a tropical ocean. The vivid greens of new spring leaves. Now look at those same colors on your monitor. They’re close—sometimes very close—but never identical. This isn’t a limitation of your particular display. It’s a mathematical impossibility baked into the physics of light and the geometry of color perception.

The Horseshoe That Broke the Triangle

The CIE 1931 chromaticity diagram is the definitive map of human color vision. Its distinctive horseshoe shape represents every chromaticity—the color quality independent of brightness—that a typical human eye can perceive. The curved edge, called the spectral locus, traces pure monochromatic colors from deep violet at 380 nanometers through blue, cyan, green, yellow, orange, and red at 700 nanometers. The straight bottom edge connecting the endpoints is the line of purples—saturated purples and magentas that exist only as mixtures, never as single wavelengths.

Here’s the problem: every additive color display creates colors by mixing three primary light sources. On a chromaticity diagram, three primaries define a triangle. Every color the display can produce lies inside that triangle. Colors outside the triangle are impossible to display, no matter how the primaries are mixed.

The horseshoe cannot be contained by any triangle whose corners lie within it. This isn’t an engineering limitation—it’s geometry. No three points within the horseshoe can define a triangle that covers all other points. The proof is in the concavity: the spectral locus curves inward at the bottom, creating a shape that no triangle can fully enclose.

Why Three Primaries? The Biology of Trichromacy

Human color vision uses three types of cone cells in the retina, conventionally called L (long-wavelength), M (medium-wavelength), and S (short-wavelength), though they’re often labeled red, green, and blue cones. Each type has a different spectral sensitivity, peaking at approximately 560 nm, 530 nm, and 420 nm respectively.

Because we have three cone types, our visual system reduces any incoming light spectrum—regardless of how many wavelengths it contains—to three signals. This is why trichromatic color mixing works at all. If humans had four cone types (as some birds do), our displays would need four primaries. If we had two (like most mammals), two would suffice.

The RGB primaries used in displays are chosen to match these cone sensitivities. Red light primarily stimulates L cones, green light stimulates M cones, and blue light stimulates S cones. By varying the intensities of these three primaries, a display can produce the same cone response as countless different spectral distributions—a phenomenon called metamerism.

But matching cone responses isn’t the same as reproducing spectral colors. A pure spectral color at 500 nm (cyan) sits on the curved edge of the chromaticity diagram. To reproduce it with RGB primaries, one of those primaries would need negative intensity—physically impossible. The mathematics of the color matching functions developed in 1931 show this clearly: for spectral colors between the primaries, at least one matching function goes negative.

The Evolution of Color Spaces: From sRGB to Rec. 2020

Not all triangles are created equal. The choice of primary locations determines which colors a display can and cannot show. The evolution of standard color spaces reflects a gradual expansion of these triangles.

sRGB and Rec. 709: The Baseline

The sRGB color space, standardized by HP and Microsoft in 1996, uses the same primaries as Rec. 709 (the HDTV standard from 1990). In the CIE 1931 diagram, this triangle covers only about 35.9% of the visible chromaticities. It was designed for CRT monitors, whose phosphor technology dictated practical primary locations.

The consequences are visible everywhere. Saturated cyans, deep greens, and vivid oranges fall outside the sRGB triangle. A photographer editing images from a high-end camera can’t accurately see colors that the camera captured but the monitor can’t display.

DCI-P3: Cinema Colors Come Home

The DCI-P3 color space, developed by Digital Cinema Initiatives in 2005, expands the triangle significantly. Its green primary is much more saturated than sRGB’s, and its red primary is a deeper hue. DCI-P3 covers about 53.6% of visible chromaticities—a 50% improvement over sRGB.

This color space was designed for digital cinema projectors, where xenon arc lamps provided enough brightness to render these more saturated primaries. By 2015, it had migrated to consumer devices, with certain smartphones and computers featuring “wide gamut” P3 displays.

Rec. 2020: The Future Standard

The Rec. 2020 color space, defined by the ITU in 2012 for ultra-high-definition television, uses primaries that lie almost on the spectral locus itself—effectively monochromatic red at 630 nm, green at 532 nm, and blue at 467 nm. This triangle covers 75.8% of visible chromaticities.

Rec. 2020 still misses about a quarter of visible colors—including some saturated cyans and deep greens. Even with primaries pushed to the edge of the spectral locus, the triangle cannot fill the horseshoe. Laser-based displays can approach Rec. 2020 coverage, with some laboratory demonstrations reaching 97% or more, but full coverage remains impossible.

xychart-beta
    title "Color Space Coverage of Visible Chromaticities"
    x-axis ["sRGB", "Adobe RGB", "DCI-P3", "Rec. 2020"]
    y-axis "Coverage (%)" 0 --> 100
    bar [35.9, 52.1, 53.6, 75.8]

The Physics Behind Display Limitations

Understanding why displays can’t reach further requires examining how they actually produce color.

LCD Technology: Filtering White Light

Liquid crystal displays don’t emit colored light directly. A white backlight—traditionally a cold-cathode fluorescent lamp (CCFL), now typically white LEDs—shines through a layer of liquid crystals that selectively block light. Color filters then split this white light into red, green, and blue components.

The color filters are the critical limitation. A perfect red filter would transmit only wavelengths around 630 nm and block everything else. Real filters have broader transmission spectra—they let through some green and blue as well. This means the red primary isn’t truly monochromatic; it’s a mixture that appears red but has lower saturation.

Quantum dot technology, recognized with the 2023 Nobel Prize in Chemistry, has improved this. Quantum dots are semiconductor nanocrystals that emit narrow-band light when excited by blue or ultraviolet illumination. They can produce primaries much closer to the spectral locus, enabling displays that cover over 90% of Rec. 2020.

OLED Technology: Organic Emitters

Organic light-emitting diodes don’t need backlights or filters. Each subpixel emits its own light through electroluminescence. In principle, this should allow more saturated primaries—you’re not filtering a broad spectrum, but generating light at specific wavelengths.

In practice, OLED primaries are also limited. The organic compounds used have emission spectra determined by their molecular structure, not arbitrarily chosen wavelengths. Red OLEDs typically emit around 610-620 nm, green around 520-540 nm, and blue around 450-460 nm. While this allows excellent coverage of DCI-P3, reaching Rec. 2020 requires more saturated primaries that current materials don’t efficiently provide.

The Pointer’s Gamut: Real Colors Matter More

The CIE diagram shows all perceivable colors, but not all are equally relevant. Many colors outside Pointer’s gamut—the set of colors that can be reflected by real surfaces—are only achievable through direct emission. The sky at sunset, neon signs, and laser light all fall outside Pointer’s gamut, but most photographed subjects don’t.

Pointer’s gamut, derived from Michael R. Pointer’s 1980 research, covers about 47.9% of the CIE 1931 chromaticity diagram. Crucially, its shape is irregular—it bulges into regions that RGB triangles struggle to cover. sRGB covers only 69.4% of Pointer’s gamut. DCI-P3 covers 86.9%. Rec. 2020 covers about 99%.

For practical photography and video, covering Pointer’s gamut should be sufficient. Rec. 2020 nearly achieves this, making it a reasonable target for consumer displays. But for scientific visualization, artistic expression, or accurate reproduction of emissive phenomena, the remaining gaps matter.

Why Not More Primaries?

If three primaries can’t cover the visible gamut, why not use four, five, or more? Some printing processes use six or more inks. Multi-primary displays have been demonstrated in laboratories.

The answer lies in engineering complexity and diminishing returns. Each additional primary requires a separate subpixel, reducing resolution for a given pixel density. It also requires modifications to image encoding, graphics pipelines, and display controllers—substantial changes to infrastructure optimized for three-primary systems.

More importantly, the relationship is non-linear. Adding a fourth primary between the red and green of Rec. 2020 would cover additional colors, but the gains would be modest compared to the jump from sRGB to DCI-P3. The horseshoe’s shape means that even four or five primaries can’t achieve complete coverage—the convex hull of any finite set of points within the horseshoe will always leave regions uncovered.

The Perceptual Dimension

Color perception isn’t just about chromaticity. The CIE 1931 diagram ignores luminance, making it a 2D projection of a 3D color space. A display’s color volume—the full 3D gamut accounting for brightness—reveals additional limitations.

At low luminance, displays can produce highly saturated colors. At high luminance, practical constraints reduce saturation. The phosphors or LEDs must produce sufficient light, and brighter primaries often have broader emission spectra. This is why the most vivid colors on any display are typically at mid-brightness, not maximum brightness.

The CIE 1931 diagram also lacks perceptual uniformity. Distances on the diagram don’t correspond to perceived color differences—MacAdam ellipses showing just-noticeable differences vary dramatically in size across the diagram. A display covering 70% of the chromaticity area might actually reproduce more than 70% of perceptually distinct colors, depending on which 70% it covers.

What This Means for Content Creators

Understanding color gamut limitations changes how we should think about color-critical work. A photographer using a wide-gamut display in DCI-P3 or Adobe RGB might see colors that won’t display correctly on an sRGB monitor. A filmmaker mastering for Rec. 2020 must accept that some theatrical projectors and home displays can’t show the full gamut.

Color management systems handle this through gamut mapping—converting out-of-gamut colors to the nearest reproducible color. The choice of mapping algorithm significantly affects the result. Perceptual mapping compresses the entire gamut to preserve relationships between colors. Relative colorimetric mapping clips out-of-gamut colors to the nearest in-gamut point, preserving in-gamut accuracy at the cost of detail in saturated regions.

The best practice remains working in a color space appropriate for the output medium, with an understanding of what will be lost in translation. A landscape photographer should know that the most saturated greens and cyans in foliage and sky will never display perfectly, regardless of equipment. A product photographer documenting items with specific Pantone colors should verify that those colors fall within their working gamut.

The Unsolvable Problem

In 1931, the scientists creating the CIE color spaces understood what they were formalizing. The RGB color matching functions they derived from Wright and Guild’s experiments showed negative values for spectral colors between the primaries. They created the XYZ color space to avoid negative numbers, using “imaginary” primaries outside the visible gamut.

This mathematical convenience obscured a fundamental truth: additive color mixing with real primaries cannot reproduce all visible colors. Every display technology—from CRTs to LCDs to OLEDs to the latest microLEDs—works within this constraint. The triangles get larger, the coverage percentages improve, but the horseshoe remains unfilled.

The next time you notice that a sunset on your screen looks slightly different from the real thing, or that certain colors seem to lack the saturation you remember, you’re observing not a flaw but a mathematical certainty. Your monitor is showing you a shadow of the visual world—a remarkably accurate shadow, engineered over decades of progress, but a shadow nonetheless. The full spectrum of visible light remains forever beyond the reach of any three-primary display.

References

  1. Wright, W. D. (1928-1929). A re-determination of the trichromatic coefficients of the spectral colours. Transactions of the Optical Society, 30(4), 141-164.

  2. Guild, J. (1931). The colorimetric properties of the spectrum. Philosophical Transactions of the Royal Society of London. Series A, 230, 149-187.

  3. International Commission on Illumination. (1931). Proceedings of the Eighth Session. Cambridge, England.

  4. Pointer, M. R. (1980). The gamut of real surface colours. Color Research & Application, 5(3), 145-164.

  5. International Telecommunication Union. (2012). Recommendation ITU-R BT.2020-1: Parameter values for ultra-high definition television systems for production and international programme exchange.

  6. Digital Cinema Initiatives, LLC. (2005). Digital Cinema System Specification Version 1.0.

  7. Anderson, M., et al. (1996). Proposal for a Standard Default Color Space for the Internet—sRGB. IS&T and SID 4th Color Imaging Conference, 238-246.

  8. MacAdam, D. L. (1942). Visual sensitivities to color differences in daylight. Journal of the Optical Society of America, 32(5), 247-274.