30

The Standard Observer: 2° vs 10° and the Average Eye

Every color number - every ΔE, every sRGB triplet, every printed spec - traces back to one surprising object: an average eye stitched together from about seventeen people in the 1920s. The CIE standard observer is a mathematical stand-in for human vision, and it carries strange baggage: primaries that go negative, two different versions for two field sizes, and the quiet fact that no real person is exactly it. This is the interactive guide to whose eye colorimetry actually uses.

Colorimetry · 76 4 Live Demos ~31 min read Foundations of CIE
1931
The 2° observer
1964
The 10° observer
x̄ ȳ z̄
Color matching functions
~17
People in the average
01

Whose eye is it?

Color science needs a reference eye. Your cones and mine differ a little - in number, in peak sensitivity, in the pigment that tints the central retina - so if colorimetry keyed off any one person, it would only be right for them. The CIE's answer, in 1931, was to build an average: pool careful color-matching data from a small panel of observers and define a single, idealized standard observer that everyone would compute against.

That average is expressed as three color matching functions, written x̄(λ), ȳ(λ), z̄(λ). Integrate any spectrum against them and you get three numbers - the tristimulus values X, Y, Z - that pin down the color the average observer sees. Every color space you know, from sRGB to Lab, is a transform of that XYZ. The whole edifice rests on this borrowed, averaged eye.

It is a fiction, and that's the point. No real person is the standard observer - it's a statistical construction. But because everyone agrees to use the same fiction, a color measured in Tokyo and a color measured in Toronto can be compared with confidence. The standard observer is less a description of vision than a treaty about it.
02

The color-matching experiment

Here is how the functions were measured. An observer sees a split field: a test light on one side, and on the other three primary lights (a red, a green, a blue) they can dial up and down. The task: match the test by mixing the primaries. For most test colors it works - but for pure spectral colors, especially cyans, no positive mix of the three primaries is vivid enough. The trick was to move one primary over to the test side - a negative amount - and match that. Slide the test wavelength and watch a primary go negative.

Interactive 01 · The matching experiment

Match a spectral test with three primaries

Pick a monochromatic test wavelength. The bars show how much of each primary (R 700, G 546, B 436 nm) the average observer needs to match it. When a bar drops below zero, that primary can't be added to the mix - it must be shone on the test instead. Those negative amounts, plotted across all wavelengths, are literally the color matching functions.

03

The three functions

To avoid the awkward negatives, the CIE transformed the measured red-green-blue matching data into three all-positive functions - , ȳ, - built on imaginary primaries chosen so the numbers stay tidy. ȳ was deliberately set equal to the eye's luminous efficiency, so Y alone gives luminance. These curves are the average observer. Trace a wavelength and read the three responses that become its X, Y, Z.

Interactive 02 · Color matching functions

The shape of the average eye

The three positive color matching functions of the 2° standard observer. Move the marker to any wavelength: the three heights are that wavelength's contribution to X, Y, and Z. A whole spectrum's color is just its power multiplied by these curves and summed - three numbers from a world of wavelengths.

04

2° vs 10°: field size

The 1931 experiment used a tiny 2° field - about a thumbnail at arm's length - seen by the central fovea. But the fovea is covered by yellow macular pigment that absorbs blue light, so the 2° observer is a little blue-blind. In 1964 the CIE added a 10° observer from a larger field that spills past the macula, making it more sensitive to blue. For anything bigger than a small swatch, 10° is the better match. See the same light two ways.

Interactive 03 · Field size & macular pigment

The same light at 2° and 10°

A light source rendered as the 2° observer sees it (through foveal macular pigment) and as the 10° observer sees it (larger field, less macular absorption of blue). Add blue to the source and the two observers diverge most - the macular pigment is exactly why a small patch and a large wall of the same paint can look subtly different.

05

Observer metamerism

Two lights with different spectra can match perfectly for the standard observer - a metameric pair. But the standard observer is an average, and you are not the average. Shift the cone peaks a little - as they genuinely vary from person to person - and a pair that matches for the standard observer pulls apart for the individual. This observer metamerism is why two people can honestly disagree about whether two samples match, and it's worst for spiky LED and screen spectra. Shift the observer and break the match.

Interactive 04 · Observer metamerism

A match that only holds for the average

Two colors - a smooth spectrum and a three-primary mix - are set to match exactly for the standard observer (they look identical at the left setting). Shift the observer's cone sensitivities to model an individual, and the two swatches diverge: the match was only ever true for the average eye. This is the deep reason color specs name the observer.

06

Which observer to use

The choice of observer is part of a color specification, not an afterthought. A quick guide:

2° (CIE 1931)
The default for small color areas and much legacy work - screen pixels, small patches, most historical data. Still the most common.
10° (CIE 1964)
Recommended for larger areas - paint walls, textiles, big prints - where the color fills more than a couple of degrees of vision.
Always state it
"L*a*b* under D50" is incomplete without the observer. A 2° and a 10° value for the same sample differ, especially in blue.
Don't mix them
Comparing a 2° measurement to a 10° one is comparing two different eyes. Keep the observer constant across a comparison.
The 2006 observer
A newer, physiologically-derived set of functions (CIE 170-2) exists and is age- and field-tunable, but 2° and 10° remain the industry standards.
Observer, not illuminant
Field size (observer) and light source (illuminant) are separate axes. Both must be fixed for a color number to mean anything.
07

Best practices and pitfalls

Specify observer + illuminant
Every color value needs both - e.g. "CIELAB, D50, 2°". Leaving either out makes the number ambiguous.
Match field size to observer
Use 10° for large samples, 2° for small ones - the observer should reflect how big the color actually appears.
Expect real disagreement
If two people argue over a match under LED light, they may both be right - observer metamerism is real, not carelessness.
Beware spiky spectra
Narrowband LED and display colors amplify observer differences. Broad, smooth spectra are far more robust across observers.
Don't over-trust one eye
A single person's visual match isn't the standard observer. Use an instrument for anything that has to be repeatable.
Remember it's an average
The standard observer predicts the typical response, not everyone's. Accessibility and edge cases need more than the average eye.
"There is no standard observer walking the earth. It's an average of a handful of eyes from a century ago - and yet every color you measure, buy, or ship is quietly judged by it." Editorial summary · the eye that isn't there
The takeaway: the standard observer turns a spectrum into three numbers via the color matching functions. It comes in a 2° and a 10° flavor for different field sizes, it's an average no one exactly is, and naming it (plus the illuminant) is what makes a color value mean the same thing to everyone.
08

Test your understanding

Six questions on the standard observer, color matching, field size, and observer metamerism. Instant feedback, no scores recorded - a wrong answer comes with a short explanation.

Quick check

Loading…
 
Question 1 of 6
09

Continue your journey

The standard observer is the root of the CIE system - here's what grows from it and what feeds into it.