The claim that human vision can distinguish roughly 10 million colors has circulated through textbooks, display marketing, and imaging standards for decades. That number traces not to a single definitive experiment but to mid-20th-century lab measurements involving a handful of observers, controlled daylight, and specific chromaticity diagrams. The gap between those controlled conditions and the way people actually see color on modern screens and under varied lighting raises a direct question: does the 10 million figure still hold up?
Why the 10 million color claim faces fresh scrutiny
Display manufacturers and imaging software developers routinely invoke the 10 million benchmark when designing color gamuts for phones, monitors, and cameras. Yet the experimental foundation for that number was built under narrow conditions that bear little resemblance to how most people encounter color. The conventionally quoted range of a million to 10 million discriminable colors rests mainly on datasets collected by MacAdam and Brown, according to a peer-reviewed review of colour-order systems. Those datasets recorded discrimination thresholds for individual observers viewing small color patches under fixed illumination, not the broad, shifting spectra of LED backlights or multispectral lighting rigs common today.
A straightforward hypothesis follows from that mismatch. If researchers re-ran discrimination experiments using modern multispectral lighting, wider observer panels, and strict control over luminance steps, the resulting count of distinguishable colors would likely land in a narrower range than the traditional 10 million figure. The original experiments often held luminance constant or varied it in limited ways. Once brightness differences are tightly locked down, fewer distinct color experiences emerge, because much of what people call “different colors” in everyday life involves brightness shifts rather than pure hue or saturation changes.
MacAdam ellipses and the data behind the estimate
The single most influential dataset behind the 10 million claim comes from a 1942 paper by David MacAdam, published in the Journal of the Optical Society of America. MacAdam recorded tens of thousands of color matches made by an observer under specified daylight viewing conditions. He plotted the results as ellipses on the 1931 I.C.I. chromaticity diagram, each ellipse representing the zone within which the observer could not reliably tell two colors apart. These “MacAdam ellipses” became the standard unit for expressing human color-discrimination thresholds.
Later work extended those thresholds beyond constant-luminance chromaticity differences. A study on combined differences in chromaticity and luminance showed that the total count of distinguishable colors shifts materially depending on whether brightness variation is included. When luminance steps are added, the estimated number of discriminable colors can climb; when they are excluded or tightly controlled, it drops. That sensitivity to experimental design is one reason the popular figure spans such a wide range, from one million to 10 million, rather than settling on a single agreed-upon number.
A separate peer-reviewed paper traced the widely cited estimate of 10 million surface colors to earlier technical references, confirming that the number is a citation claim passed along through successive publications rather than the direct output of any single experiment. The figure accumulated authority through repetition, not through independent replication under varied conditions.
Gaps in the experimental record
Several unresolved problems limit confidence in the 10 million estimate. No primary dataset in the published record measures discrimination thresholds across modern wide-gamut displays or under the LED and OLED spectra that dominate current screens. MacAdam’s 1942 data relied on a single trained observer. Brown’s later extensions broadened the scope but still used small observer pools and fixed illuminants. Biomedical overviews of color perception, such as those hosted by the U.S. medical library, summarize wavelength discrimination methods and relate them to hue, saturation, and brightness judgments, but they do not supply raw threshold data for luminance-inclusive conditions under contemporary adaptation states.
The absence of large-scale replication matters for practical reasons. Screen designers, color scientists working on medical imaging, and autonomous-vehicle engineers all rely on assumptions about how many colors a human operator can distinguish. If the real number under typical indoor LED lighting is closer to one million than 10 million, color-critical applications may be over-engineering gamut width while under-investing in luminance precision and adaptation modeling.
Another gap involves observer variability. The original experiments used highly trained participants whose performance may not represent the broader population, including people with mild color-vision deficiencies that go undiagnosed. Population-level discrimination data collected under standardized but realistic lighting would provide a far more useful benchmark for display and imaging engineers. It would also clarify how much headroom current standards should reserve for observers with atypical color responses.
What the color count means for screens and standards
The practical stakes extend beyond academic debate. Modern display specifications often tout coverage of standardized color spaces such as sRGB, Adobe RGB, or DCI-P3. These gamuts describe the range of chromaticities a device can render, not the number of distinct colors a person can perceive on that device. Still, marketing claims about “billions of colors” implicitly lean on the idea that human vision can resolve fine gradations across those gamuts.
If human discrimination thresholds under realistic conditions support fewer unique colors than assumed, engineers might be better served by optimizing other aspects of image quality. For example, smoother tone mapping in dark regions, more accurate white balance under mixed lighting, and improved handling of local contrast can all have greater perceptual impact than extending gamut into regions of color space that few observers can distinguish. In that scenario, the 10 million figure becomes less a hard limit and more a reminder that perceptual resolution is finite.
On the other hand, if updated experiments confirm that human observers can reliably distinguish color steps near the upper end of current estimates, that would strengthen the case for wide-gamut workflows in cinema, virtual reality, and professional imaging. It would also support efforts to standardize high-dynamic-range formats that preserve subtle chromatic differences at both very low and very high luminance levels. Either outcome would give standards bodies firmer ground than the inherited 20th-century estimates now in circulation.
Designing better experiments for the LED era
Closing the gap between legacy estimates and present-day viewing conditions will require carefully designed experiments. Researchers would need to recruit large, demographically diverse panels, including participants with varying levels of color acuity. Test stimuli should be presented on calibrated wide-gamut displays and under controlled ambient lighting that reflects common environments such as offices, living rooms, and outdoor shade.
Crucially, luminance steps must be handled explicitly. Rather than assuming constant brightness, experiments could map discrimination thresholds across a grid of both chromaticity and luminance values, using adaptive psychophysical methods to home in on just-noticeable differences. The resulting data could then be translated into updated discrimination volumes in modern color spaces used by industry, such as CIELAB or more recent perceptually uniform models.
Such work would not yield a single magic number. Instead, it would likely produce a family of estimates: how many colors can be discriminated under dim indoor lighting versus bright daylight, how thresholds change with age, and how much variation exists across observers. These distributions would give engineers and standards committees a more realistic basis for deciding how much color resolution to target in different applications.
A moving target, not a fixed constant
The enduring appeal of the “10 million colors” claim lies in its simplicity. It compresses a complex, context-dependent property of human vision into a single, memorable figure. But as lighting technologies, display hardware, and viewing habits evolve, that simplicity becomes a liability. The original experiments were never meant to define a universal constant of perception; they were snapshots of performance under specific, tightly controlled conditions.
Revisiting those estimates with modern tools would not just tidy up a long-standing number in textbooks. It would clarify how human vision interacts with the digital images that now mediate so much of daily life, from medical diagnostics to entertainment. Whether the final tally comes in closer to one million, 10 million, or a range that depends on context, bringing the science in line with contemporary conditions would help ensure that the colors our devices can display are matched thoughtfully to the colors we can actually see.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
