Evening light and your body clock: what a screen can and can't do

This page explains the one number Atapa shows about light, how it is computed, and where it stops. It is written by the people who make Atapa, and it makes no claim about your sleep.

What is melanopic light, in plain words?

Your retina has a small set of cells that do not help you see shapes. They measure how bright the world is and report it to the body clock. Their pigment, melanopsin, is most sensitive to light around 480 nm, in the blue, and much less sensitive to red. Light weighted by that sensitivity is melanopic light. CIE S 026:2018 is the international standard that defines the weighting, and Lucas et al. 2014 is the paper that argued for measuring light this way instead of by lux alone.

So the question for a screen in the evening is not “how much blue light does it block” but “how much melanopic light does it emit”. The first has no shared method; the second has a standard.

Why blue is about 72% and green about 28%

A white screen is three primaries added together. To know how much of its melanopic output each channel carries, you weight each primary’s spectrum by the CIE S 026 melanopic function and by the primary’s share of white luminance.

For an Apple-style P3 panel we modelled a 450 nm blue LED, a 533 nm green phosphor and KSF red lines at 613, 631 and 648 nm, with P3 luminance shares of 0.229, 0.692 and 0.079. The melanopic-to-luminance ratios come out at 0.0023 for red, 0.614 for green and 13.57 for blue, and the channel shares at:

Broader P3 primaries give 0.000, 0.313 and 0.686. An sRGB white-LED monitor gives 0.004, 0.252 and 0.743. The split moves a few points between displays, which is why the app always says “about” and “estimate”. It replaced an earlier guess that weighted green at 10% and overstated the effect; green matters more than the “blue light” framing suggests.

Why dimming matters as much as warmth

Warming a screen removes blue and green. Dimming it removes everything in proportion. Melanopic light scales with luminance, so the same spectrum at half the brightness is half the melanopic output. The two effects multiply: a warm, dim screen emits far less than a warm, bright one.

That is why Atapa’s evening curve dims as it warms, and why it can dim below the lowest brightness step of the hardware. It is also why a mild orange at full brightness, which is what Night Shift gives at More Warm, still leaves a lot of melanopic light on the screen.

What the ring estimates

Atapa’s popover shows one ring: “about x% less alerting light”. It is the melanopic estimate for the current tint against an untouched screen, from the linear output multipliers for warmth, Readable red and dimming. In Statistics it is tracked per evening, from 18:00 to 06:00; time with the tint off counts as 100%, and a sleeping display is not counted.

At 100% brightness, Atapa’s own color math gives these reductions. Each value is simulated with Atapa’s own color math on the P3 panel model, not measured on your display.

Dimming multiplies the remainder. At 2700K and 90% brightness, the remainder is 0.20 times 0.9, so the ring reads about 82% less. At 500K with Readable red the estimate reaches about 100%, because the green and blue outputs are zero; the small red output carries almost no melanopic weight. Night Shift’s warmest step, around 2700 to 3000K by independent measurement, lands near the 74 to 80% rows at full brightness.

What the research does not show

The papers below show that melanopsin drives the body clock, that its sensitivity peaks in the blue, and that evening light should be dim and low in melanopic content. Brown et al. 2022 give a recommendation for the three hours before bed in melanopic terms. Cajochen et al. 2011 found that an LED-backlit screen in the evening affected circadian physiology and alertness compared with a less blue screen. Spitschan 2019 reviews what melanopsin does and does not contribute.

None of that is a study of Atapa, and we have not run one. We do not claim that Atapa improves sleep, raises melatonin, or is medically proven. We do not publish a “blue light blocked” percentage, because the figures on vendor sites share no method. We do not claim to remove PWM flicker: the backlight of a MacBook Pro flickers in hardware, and no software changes that. What we claim is narrower: Atapa reduces the screen’s melanopic output, by about the amounts in the table, as an estimate.

Sources, and what each one says

1. CIE S 026:2018 defines the melanopic weighting function and the five alpha-opic quantities. It is the standard behind the 72/28/0 split.
2. Lucas et al. 2014, Trends in Neurosciences, argues for measuring light by its melanopic content rather than by lux alone, and supplies the toolbox that later work uses.
3. Brown et al. 2022, PLOS Biology, is a consensus recommendation for indoor light across the day: bright and melanopic-rich in the daytime, dim and low in melanopic content in the evening and at night.
4. Cajochen et al. 2011, Journal of Applied Physiology, is an evening-screen study: an LED-backlit screen with more blue affected circadian physiology and cognitive performance compared with a less blue screen.
5. Spitschan 2019, Current Opinion in Behavioral Sciences, reviews melanopsin’s contributions to both non-visual and visual function and where the evidence is still open.