Color Theory/Misconceptions

From Wikibooks, open books for an open world
Jump to navigation Jump to search

Due to easy access to colors, especially in the digital era, and lack of an appropriate education, there are numerous misconceptions about them. Some are (explicated) wrong statements and beliefs, but many misconceptions constitute a misunderstanding of certain core concepts which damage abilities of a person to operate with colors and describe them correctly.

This chapter does not consider deeply ignorant, childish misconceptions such as “there exist only (whatever small number) colors” and “all colors are RGB”.

Universality of perception[edit | edit source]

There are two opposite extreme views:
Colors are always the same for most people in most circumstances. The perception of colors depends on numerous factors including illumination, duration, surrounding, angle of view, a human’s genome, psychological factors, and cultural background of a person.

Technically, there are two levels of color perception: the low level (photoreceptors in the retina), and the high level (further processing by neurons). The low level is actually almost universal (except noted below) and is formalized in color spaces which are usable for the great majority of humans. The high level really depends on numerous factors.

Now, let’s comment concrete points:

  • Illumination: There are some effects of brightness on low-level perception, but they can usually be neglected. Chromatic adaptation, a high-level effect, is important, though.
  • Duration: I don’t know.
  • Surrounding: Yes, important for high-level perception.
  • Angle of view: Yes, it affects low-level perception because most cone receptors are concentrated around the center of visual field.
  • Human’s genome: For about 80% of humans the low level of color perception is nearly identical.
  • Psychological factors and cultural background: Yes, for the high level. These do not affect colors discretion ability, but affect colors memorizing.
Conclusion: colors are objective, but are not always perceived identically.

Color terms[edit | edit source]

Color terms, at least basic color terms such as “black”, “white”, “red”, “green”, “blue”, and “yellow”, admit a universal and natural definition.

Actually, only black, which is defined as darkness, i.e. absence of visible light. No so for any other color term. The only thing about colors which admits a universal and natural definition, is spectral colors. But there are many. It is not true that only 7 spectral colors (red, orange, yellow, green, blue, indigo, violet) exist. Physically, their number is not limited, and human’s eye is able do discreet hundreds. Frequently humans think (or feel) that there are several primaries of which other colors (including most spectral ones) are mixes. Although it is true that a spectral color looks nearly identical to a mix of two nearby, flanking colors (for example, spectral lime would not be distinguishable from appropriate mix of spectral yellow and green), no color but two spectral extremities is exceptional in it. Likewise, yellow created from mixing of spectral orange with spectral lime would not be distinguished from spectral yellow. Some researches indicated that some spectral colors are perceived as pure colors psychologically: see w:Unique hues. But same researches confirmed a significant discrepancy in definitions of “unique hues” made by different individuals. Note that the color vision has almost identical physiological parameters for 80% of humans. These “primaries” reflect psychological processes and cultural prejudices of an individual. To make the matter worse, meaning of some color terms in some languages evolved over time. For example, computer’s standard blue was known as “indigo” by Newton, and the “blue” of 19-century writers extended sometimes as far towards green as up to cyan.

The only two colors which obviously cannot be created as a mix of others, and hence should possess certain natural excellence, are extreme spectral red and extreme spectral violet. But the latter, although present in natural light, perceived by a human eye dimly and does not play a prominent rôle in the world of colors. So, only red, but it has its special problems and anyone can never be sure that another person’s “red” is spectral. Finally, see #Light, pigments, and universal color classification about problems with definition of white.

Along “black”, the only term of these six (relatively) spared of historical and technical discrepancies is “yellow”. Hues of traditional yellow, of RGB’s secondary yellow = red + green, and CMYK’s primary yellow do not differ greatly.

Conclusion: there are no naturally defined color terms but black.

Spectrality of red[edit | edit source]

Red is a spectral color.

Yes, a rather natural definition of red does exist: the color of extreme low-frequency segment of the visible spectrum; this hue is known as carmine. But for a number of reasons it is not a standard red. First, it is not a traditional definition. Red was historically the first color term proper, which appeared in human languages just after white and black. Traditionally, it encompassed a wide range of color, including non-spectral ones, such as crimson, rose, and even magenta. Actually, only that side of the traditional “red” which borders orange has any dominant wavelength, (i.e. a hue compatible with spectrality). Red is a possibly spectral color, just like other color term. Or: there is a part of spectrum which is red.

Even worse, the standard modern red (a primary color of sRGB) is not far from spectral red, but actually is classified as… reddish orange. It lies on the orange side of red/orange boundary of Newton and Goethe (see below), and definitely does not represent what is known as red by spectroscopists. Unlike spectral orange, middle spectral red is far from being accessible in sRGB.

Conclusion: red can be a spectral color, some of its variations, but they are rarely seen in practice.

Light, pigments, and universal color classification[edit | edit source]

There are two opposite views:
There exists one-to-one correspondence between colors of light and colors of pigments (things which do not emit their own light). Hence, a universal color dictionary is possible. The correspondence between the color of an object (not light-emitting) and chromaticity and lightness of its scattered light is complicated. Hence, a universal color dictionary cannot exist.

An ideally white thing reflects (diffuses) all incident light. But which exactly light: of the Sun? Diffuse sky radiation? An incandescent lamp? They all have different spectra and perceived colors. So, one ideal white can correspond to many reflected colors of light: see w:White point for further explanations. Also, pigments which look identical under a bright sky can become different under incandescent lamp, and vice versa. But typical materials experience a regular, predictable shift of diffused light’s chromaticity depending on illumination. This shift is not large near spectral colors. Even when shift is noticeable, it does not result in dramatical change of hue of most pigments under usual illuminants. A violet/purple (on daylight) clothing will become (objectively) carmine under a red lamp, but it will not reflect carmine (or even crimson) if illuminated by unfiltered incandescent lamp at ~ 2500 K, but would become something like electric magenta.

Conclusion: a universal, but not precise, description of colors is possible.

Color space representation[edit | edit source]

Essentially, one misconception, but it may have stronger and weaker forms:

Any color can be produced from mixing of 3, 4, 5, 6, or 7 (depending on definition of “mixing” and amount of knowledge) primary colors.

Weaker: The space of all colors is described with a simple (3-dimensional) shape.

A stronger version implies that this shape would be a convex polytope. No, all colors do not form neither a triangle (or a pyramid), hexagon, cube, nor a cone, ball, or something alike. It was explained above that a small number of magically pure colors, which could form vertices of this shape, does not exist. So polygons, pyramids, and cubes which include all colors can be ruled out. Actually, there exist a 3-dimensional shape which describe the space of all possible colors, but it is not simple. See w:color space for details.

Conclusion: there are no geometrically simple color spaces.

Complementary colors[edit | edit source]

There exist certain “RYB complementary” (variants: “subtractive complementary” or “traditional color model”), which results in rather different color pairs than exist in the standard (additive color) complement. For example, “RYB/subtractive complementary” of red is green.

This misconception is notorious for dwelling even in minds of some (near-)experts. Theoretically, “subtractive complementary” is different from the additive one, because rules for color mixing do not coincide. Moreover, mixing of pigments can produce different results depending on their absorption spectra. But actually… not so.

Many authors refers to works of Goethe to substantiate an existence of certain “traditional color model”. There were many such models in 19th century, and Goethe’s one (the earliest) is not so different from modern views. Goethe described the complementary of orange as Blau (blue). What says the modern theory about the complementary of orange? It is azure, a color between the modern color wheel blue and cyan. But what did represent Goethe’s Blau? Indeed, it is the same as azure, both modern azure and traditional azure from heraldry. The term Blau (blue) referred to a wide range of hues from the modern standard blue to cyan. It was centered roughly at what is now called azure, the color of sky. Do you see the computer’s standard blue on the famous self-portrait of van Gogh? The difference lies in terminology, not in an actual palette. The same about the complementary of yellow: the computer’s standard blue, a.k.a. Newton’s “indigo”, does not differ greatly from Goethe’s “violet”.

Finally, consider the question of red–yellow–blue “color model” and alleged red–green “complementarity”. This also is largely a question of color wheel terminology. Contrary to some interpretations, Goethe’s “Theory of Colors” does not say anything about the complementary of red. But it says that the complementary of green is… Purpur. In modern language, magenta—the same pair of colors as the modern color theory has. What Goethe actually did is arrangement of colors into six sectors. Purpur and red (Rot) inhabited the same sector (called schön), but Goethe never said they are the same. Goethe wrote that blue and yellow are special, but did Goethe write that red is special? Red–yellow–blue triad was a later 19th-century invention, and it actually lacked a proper definition of “red”, and also a proper definition of “blue”. Variants of blue which mix with yellow and give saturated colors differ from variants of blue which mix with red and give saturated colors. Naming of colors as “blue” or “red”, and formulae like violet = red + blue are rather arbitrary and lack a precision. Some non-spectral variants of traditional “red” are actually complementary to some definitions of green (such as Munsell and rather modern NCS), which perfectly can be produced as a mix of yellow with certain “blue”. And physical difference between additive and subtractive complementaries certainly has nothing to do with the case of red, as subtractive complementary to CMYK’s red (the hue almost identical to standard red) is cyan, never green. So, “RYB complementaries” are actually a question of confused terminology (this “R” might be closer to magenta than to the modern standard red), it have little to do with any (either physiological or pigment mixing) color process.

Conclusion: the problem of complementary colors is aggravated by inconsistent use of color terms.

Color mixing in RGB[edit | edit source]

In RGB color space, color mixing can be made with convex combinations. E.g. the color exactly midway between RGB’s red (255, 0, 0) and RGB’s yellow (255, 255, 0) is (255, 128, 0).

Although this simplistic algorithm is sometimes used, it is incorrect, theoretically, because of strong non-linearity of most RGB implementations, such as sRGB. The “128” value actually gives only about 22% luminance compared to “255”, not 50%.

Another version of this misconception is a simplistic formula for the complementary, (255 − r, 255 − g, 255 − b). It gives correct results for primary and secondary colors, but deviates from physical complementary colors in all other hues.

Conclusion: a monitor’s RGB is not a space where a straight line can be created with linear functions.