What is color temperature?

Color temperature sits at the intersection of physics and perception. In purely physical terms, an idealized object emits light whose spectrum, and thus apparent color, depends on its temperature. Yet human experience tells a more subjective story: we tend to associate reds and oranges with warmth, while blues and crisp whites feel cool. But where does that leave colors in between, like green?

Physics: blackbody radiation

A blackbody is an ideal object that absorbs all radiation and emits light purely based on its temperature. This emission follows Planck’s law.
As temperature increases:

  • ~1000 K → barely visible, very dim dull red
  • ~1200–1500 K → visible deep red (heated metal just glowing)
  • ~2000 K → orange-red
  • ~2700–3000 K → warm white (incandescent bulbs)
  • ~4000–5000 K → neutral white
  • ~6500 K → daylight white
  • ~8000–10000 K → cool white with slight blue tint

This continuous spectral output is crucial: it contains all wavelengths, just distributed differently.

2140474karishma, CC BY-SA 4.0, via Wikimedia Commons

Map physics into chromaticity space

Humans don’t perceive light as linear wavelengths; instead, the cones in our eyes convert incoming signals into three channels (LMS). To represent this, the CIE 1931 color space maps visible colors into a 2D horseshoe-shaped diagram. When the continuous spectral output of a blackbody, described by Planck’s law, is projected into this space, it doesn’t appear as a straight line but forms a curve known as the Planckian locus, showing how perceived color changes with temperature.

The Planckian locus traces the colors emitted by blackbody radiation, giving us a physics-based way to label light as “warm” or “cool.” Physically, higher temperatures shift emission toward shorter wavelengths (bluer light), while lower temperatures correspond to longer wavelengths (redder light), as described by Planck’s law. However, human perception flips this interpretation: we tend to associate reds and oranges with warmth and blues with coolness. This contrast highlights the gap between physical reality and how we intuitively interpret color.

en:User:PAR, Public domain, via Wikimedia Commons
Color palette png sticker, art

In physics, higher temperature (Kelvin) means greater energy output and a shift toward shorter wavelengths, as described by Planck’s law, whereas in color theory this is intuitively reversed, with reds perceived as “warm” and blues as “cool.”

How we calculate temperature in color theory?

You can see how blackbody radiation gives us a physically grounded way to assign temperature—but only for the colors that lie along the Planckian locus. For all other colors, a practical and intuitive approach is to map them to the nearest point on this curve: if a color lies closest to a region associated with lower temperatures (reddish tones), we perceive it as warm, and if it is closest to higher-temperature regions (bluish tones), we perceive it as cool, effectively extending the idea of color temperature beyond the locus itself.

warm vs hot

This raises further questions: what if a given color has two equally close points on the Planckian locus, one corresponding to a warm temperature and the other to a cold one? Can we meaningfully call a color “cold” simply because its nearest point lies in the cold region, even if it is far from the locus itself? More generally, how should the distance from the locus be incorporated into a temperature assignment, and is there a principled formula that combines proximity and chromatic deviation into a single meaningful temperature value?

Read how we calculate the temperature in Negarity Color library.

Correlated color temperature (CCT)

Correlated color temperature (CCT) for a digital color (like an sRGB value on the web) is not computed by a single direct formula, but through a short pipeline that maps the color into a physically meaningful space and then compares it to ideal blackbody light. First, the sRGB values are linearized (removing gamma correction), and then converted into CIE XYZ tristimulus values using a fixed matrix. From these, chromaticity coordinates are derived as (\(x = \frac{X}{X+Y+Z}\)) and (\(y = \frac{Y}{X+Y+Z}\)). These coordinates are often further transformed into the CIE 1960 UCS space using (\(u = \frac{4x}{-2x + 12y + 3}\)) and (\(v = \frac{6y}{-2x + 12y + 3}\)), which is more perceptually uniform for temperature comparisons.

CCT is then defined as the temperature (T) of a blackbody radiator whose chromaticity lies closest to the given color in this space. In practice, this means finding the value of (T) that minimizes the distance (\( (u – u_T)^2 + (v – v_T)^2\)), where (\((u_T, v_T)\)) lies on the so-called Planckian locus (the curve of ideal thermal light sources). Because this inverse mapping cannot be solved analytically, web implementations typically use either numerical search along the locus or fast approximations such as McCamy’s formula (\(\mathrm{CCT} \approx -449n^3 + 3525n^2 – 6823n + 5520\)), where (\(n = \frac{x – 0.3320}{y – 0.1858}\)).

More information

What are those warmest and coldest possible colors?

If you look at the Planckian locus on the CIE chromaticity horseshoe, you’ll notice that the curve terminates in a distinctly bluish region. Based on experimental approximations, this endpoint is often associated with a theoretical light source of infinite spectral power. In a symbolic sense, it represents the highest possible energy state of visible color.
Since color theory frequently frames warm and cool tones as perceptual opposites, this extreme blue is sometimes described as the “coldest” imaginable color. This color is known as Perano, and you can explore perano RGB representation and related details here.
The minimum-energy color in the visible spectrum is deep spectral red, located at the longest visible wavelengths near the infrared boundary. Unlike the blue end of the Planckian locus, which approaches an asymptotic limit, the red end gradually fades out of visibility into infrared radiation. But if you want a realistic blackbody warm endpoint, you can explore rgb(255 106 26) and related details here.

Is that possible for two different visible colors to have the same temperature?

When working directly with the Planckian locus, each point on the curve corresponds to a unique correlated color temperature, so on the locus itself there is a one-to-one relationship between color and temperature. However, when we estimate temperature by mapping the full two-dimensional CIE chromaticity diagram (the “horseshoe”) onto this one-dimensional locus, that uniqueness is lost. In this projection, many different colors can collapse onto the same or very similar temperature value, since distinct chromaticities are being reduced to a single scalar measure.

Is the temperature value persistent between different color spaces?

If the proper color-space conversion is omitted and the same numerical representation is interpreted directly, the result is only consistent within that representation and not physically meaningful across systems. However, when colors are compared through a shared chromaticity model, consistency depends on whether they lie within a compatible gamut. If a color exists within the overlapping gamut of two different color spaces, then—ignoring conversion precision and numerical error—it will map to the same chromaticity and therefore yield the same color temperature under the same Planckian-locus-based method.

Is CCT the only possible estimation for color temperature? If not, what are the other alternatives?

No, CCT (Correlated Color Temperature) is just the most common engineering-friendly approximation, not the only way to define or estimate “color temperature”. There are other models such as Spectral (true physical) color temperature, Isotemperature lines / Duv-based models, and Planckian distance metrics (Δuv / ΔE).

Why blackbody radiation can’t emit all the visible colors?

A blackbody does not emit a set of independent visible colors; instead, it produces a continuous spectrum whose shape is entirely determined by its temperature. As that temperature changes, the spectral peak shifts smoothly across wavelengths, but the distribution always remains a single-peaked curve rather than a mix of controllable colors. This is why blackbody radiation traces the one-dimensional Planckian locus in chromaticity space: it can move through different perceived hues (from red to white to blue-white), but it cannot independently generate or combine all possible visible colors

References

  • https://cie.co.at/eilvterm/17-23-067
  • https://en.wikipedia.org/wiki/Color_temperature
  • https://www.handprint.com/HP/WCL/color12.html#warmcool