Color Temperature Explained: What Kelvin Means for Your Mac

Color temperature is a way of describing the color of light with a single number, measured in kelvin (K). A low number like 1900 K is the orange of a candle flame, a high number like 6500 K is the cool white of average daylight, and your Mac ships set to that daylight white whether it is noon or midnight. After dark, it is the most useful setting on your screen.

Below: where the kelvin scale comes from, why 6500 K became the standard screen white, what happens inside a display when you lower its color temperature, and what the research on evening light says, including the parts that do not flatter anyone selling a screen filter. Sundown sells one, so every study is linked.

What is color temperature?

Color temperature is the temperature, in kelvin, at which an ideal glowing object (a "blackbody") gives off light of a particular color. Heat an object and it glows dull red, then orange, then yellow-white, then bluish-white, and physics fixes exactly which color belongs to which temperature.

A blackbody absorbs everything that hits it and emits purely according to its temperature. In 1900 Max Planck worked out the formula for its spectrum, still called Planck's law, and the part that matters for your screen is simple: the hotter the object, the more of its light sits at the short, blue end of the spectrum.

The kelvin is the SI unit of temperature, counted from absolute zero (the US National Institute of Standards and Technology keeps the definition). An incandescent filament really is a hot glowing object, running in roughly that 2700 K range, but most modern lights are not: LEDs, fluorescent tubes and screens make light without being hot. So the formal term is correlated color temperature, or CCT. It describes a light's color (its chromaticity) by the blackbody temperature it most closely matches, whether or not anything in it is actually that hot.

Why is 6500 K called "daylight" on a screen?

6500 K is shorthand for CIE Standard Illuminant D65, the reference the lighting world uses for average daylight, and the sRGB and Display P3 color standards both use D65 as their white point. Your Mac's white is tuned to it so photos and video look the way their makers intended.

D65 models daylight arriving at the ground, sunlight mixed with blue sky, and its correlated color temperature works out to roughly 6504 K. That makes sense at two in the afternoon. It makes less sense at eleven at night in a room lit by one lamp, because the screen keeps emitting noon-colored light into a room that has long since gone amber.

Planck's law also tells you where a blackbody's output peaks. At 6500 K the peak falls at about 446 nanometers, in the blue. At 2700 K it sits out near 1,070 nanometers, in the infrared, which is why an incandescent bulb feels warm to the hand and looks gold to the eye. Pull the color temperature down and you slide the balance of the light away from the blue end.

What does a 2700 K screen actually look like, and why?

A screen set to 2700 K looks like a page under a bedside lamp: whites turn gold, blues turn muddy, and the display is still perfectly readable. It gets there by driving its blue subpixels much less hard, and at lower settings its green ones too.

A screen is not a blackbody and never becomes one. An LCD or OLED panel makes every color from red, green and blue primaries, and "white" is all three at chosen strengths. Software cannot change the LEDs, so to lower the color temperature it turns down the blue channel until the mix matches a warmer blackbody's white. The blue peak stays where the hardware put it; what changes is how much of it reaches you.

At the very bottom of the range the Kelvin number becomes a label rather than physics. A real object must reach roughly 800 K before it visibly glows, so a "500 K" screen describes no lamp you could buy. It means a deep red setting far below any household bulb.

Apple's Night Shift covers the upper part of this range. When researchers at the Lighting Research Center at Rensselaer Polytechnic Institute measured it on an iPad with a spectrometer, its warmest setting came out at 2837 K and its coolest at 5997 K (Nagare, Plitnick and Figueiro, 2019). If you want to push further on a Mac, our guide on making Night Mode on your Mac darker walks through the options.

Melanopic light: the part of the spectrum your body clock counts

Your body clock does not measure brightness the way your eyes do. It reads light through melanopsin, a pigment most sensitive to short-wavelength blue light, and scientists now quantify that input as "melanopic EDI", defined in the international standard CIE S 026:2018.

In 2001 two teams, George Brainard's at Thomas Jefferson University and Debra Skene's at the University of Surrey, mapped which wavelengths suppress melatonin in humans. Both found peak sensitivity in the blue, away from where rods and cones are most sensitive, which pointed to an unknown photoreceptor (PMIDs 11487664 and 11507175). In 2002 David Berson's lab at Brown University identified it: a third kind of light-sensing cell in the retina (Berson, Dunn and Takao, Science, PMID 11834835). These intrinsically photosensitive retinal ganglion cells (ipRGCs) build no images; they report ambient light to the brain's master clock. Steven Lockley, Brainard and Charles Czeisler later showed that 460 nm blue light produced a two-fold greater circadian phase delay than 555 nm green light at equal photon density (PMID 12970330).

Ordinary lux is weighted to how bright light looks to your visual system, which peaks in the green, so two lights can read the same on a lux meter and hit your clock very differently. Robert Lucas and colleagues at the University of Manchester set out the case for a melanopsin-weighted measure in 2014 (Trends in Neurosciences, PMID 24287308), and the International Commission on Illumination (CIE) formalized it in CIE S 026:2018.

That standard defines melanopic equivalent daylight illuminance, or melanopic EDI: the amount of D65 daylight, in lux, that would stimulate melanopsin as strongly as the light you are actually looking at. Spectrum decides the score, so it is not a rule for every light, but for typical broadband white sources a warmer one usually scores lower melanopic EDI than a cooler one of the same visible brightness. On a display the effect is more direct: warming it means turning down the blue channel, the part of its output melanopsin responds to most, and that is the physical case for warming a screen.

In 2022 an international panel led by Timothy Brown at the University of Manchester, with co-authors including Czeisler and Lockley of Brigham and Women's Hospital, turned the standard into targets (PLOS Biology, PMID 35298459). For healthy adults they recommend at least 250 lux melanopic EDI during the day, no more than 10 lux melanopic EDI in the evening starting at least three hours before bedtime, and under 1 lux while asleep. They are consensus recommendations, not trial results, and people differ widely in sensitivity, but they are the clearest published answer to "how dim is dim enough".

What does warming your screen do, and what does it not do?

Warming a screen lowers how much short-wavelength light it emits, which lowers its melanopic output at a given brightness. It is not a measured percentage of blue light blocked, and on its own, at full brightness, it may not be enough.

Apps and glasses often quote a precise share of blue light "blocked". A figure like that is an engineering measurement of one screen at one setting, dependent on panel, brightness and instrument. It is not a clinical outcome, and we do not quote one for Sundown because no spectroradiometer trace on file backs it.

The most direct test of a warm-screen mode is the Rensselaer study above. Twelve young adults used iPads at maximum brightness for two hours late at night, once with Night Shift at its warmest and once at its coolest. Melatonin suppression did not differ significantly between the two, and the authors concluded that changing the color of a display "without changing their brightness settings may be insufficient." The sample was small and the trend pointed the right way, so it does not show warmth is useless either.

The honest reading is that color and brightness work together. Melanopic EDI rises with both, so the sensible evening setting is a warm screen that is also dimmer. That is why Sundown keeps brightness as its own control beside warmth, and why we suggest using both.

What does the research say about evening screens and sleep?

Evening light, screens included, can delay the body clock and suppress melatonin, and blocking short wavelengths has improved sleep in small trials. The evidence is suggestive, not settled, and the largest review of blue-light glasses found little to support them.

The screen study people quote most, from Anne-Marie Chang and Charles Czeisler at Brigham and Women's Hospital (PNAS, 2015), compared a tablet with a printed book, not a warm screen with a cool one, so it tells you screens matter without telling you which color temperature helps.

The strongest blocking result is this: Ostrin et al. (2017) found blue-blocking lenses raised nocturnal melatonin ~58% in healthy adults, who also slept ~24 minutes longer. Those were 21 adults wearing glasses, not using a screen filter, and the transfer to software is a reasonable inference rather than a finding.

Against our own interest, the 2023 Cochrane review of blue-light filtering spectacle lenses concluded they probably make little or no difference to eye strain from computer use, and found no randomised-trial evidence on melatonin at all. Heo and colleagues (2017, PMID 28017916) found blue-reduced smartphone use improved sleepiness while the melatonin change was not statistically significant. For a plain-language overview from outside the industry, the Sleep Foundation's guide to blue light is a fair place to start.

Most of these studies, with what each does and does not show, sit on our Sundown science page. Three are cited here directly and are not on that page yet: Brown et al. 2022, Lucas et al. 2014 and Nagare et al. 2019.

How Sundown uses Kelvin

Sundown sets your Mac's color on a Kelvin ladder from 6500 K down to 500 K in 100 K steps, with a warmth intensity scale from 0 to 4.0 and a separate brightness control, so every setting is a number you can read, repeat and compare.

We chose Kelvin over a vague "warmth" slider because a color temperature means the same thing everywhere: 2700 K on Sundown is the color of your bedside bulb, so you can tell a friend a number instead of "about two-thirds of the way along".

A reasonable routine, based on the Brown 2022 timing rather than any study of Sundown, goes like this. Start warming about three hours before bed, somewhere around 3000 to 2700 K. Lower the brightness as the room gets darker, because brightness matters as much as color. In the last hour, if you are still working, go lower, toward 1900 K and below, which reads like candlelight. The blue light calculator gives a rough estimate of your evening exposure, and our comparison of Night Shift vs f.lux covers the built-in and free options.

No study has tested Sundown itself. What it does is move your display's white point to the color temperature you choose and hold it there, and you can also build a custom schedule by time of day. Try Sundown tonight and set your screen by the number.

Frequently asked questions

Is a higher Kelvin warmer or cooler?

Cooler. Higher Kelvin numbers look bluer and lower ones more orange: 6500 K is daylight white, 2700 K a household bulb, 1900 K a candle.

What color temperature should my Mac screen be at night?

There is no clinically tested number for screens. Consensus guidance (Brown et al., 2022) says keep melanopic light low from three hours before bed, so warm to around 2700 K or lower and dim the screen as well.

What is melanopic EDI?

Melanopic equivalent daylight illuminance measures light by how strongly it stimulates melanopsin, the pigment that signals your body clock. It is defined in CIE S 026:2018 and expressed in lux of equivalent daylight.

Does warming my screen block a set percentage of blue light?

No honest single figure exists. Warming lowers short-wavelength output, but how much depends on the panel, brightness and instrument. We do not quote a blocked percentage for Sundown because no spectral measurement supports one.