To Galleries
Home Galleries Articles Reviews Best Gear Science New About Contact

Color Vision at Night

by Roger N. Clark

The night sky and landscape illuminated by the night sky are bright enough to see color, contrary to popular opinion. Generally, two factors limit our ability to see color at night: 1) inadequate dark adaption due to artificial lights, and 2) low contrast between objects in the night sky and the sky background due to light pollution and airglow.

The Night Photography Series:


The Magnitude Scale
The Color and Brightness of the Sky and Landscape at Night
Photometry and Threshold Color Perception
Discussion and Conclusions
References and Further Reading

All images, text and data on this site are copyrighted.
They may not be used except by written permission from Roger N. Clark.
All rights reserved.

If you find the information on this site useful, please support Clarkvision and make a donation (link below).


The human visual color perception ranges from about 4000 Angstroms in the blue (400 nm, 0.4 microns) to about 7200 Angstroms in the deep red (720 nm, 0.72 microns). See "What Wavelength Goes With a Color?" for more information. As light levels fall, our color perception decreases and we see only grey when the scene becomes very dark. At bright light levels, e.g. daytime to typical indoor color, people with normal vision can perceive a huge range of colors and colors can appear quite saturated. The brightness range where we see full color is called photopic vision and we see color using the cone cells in our retinas. At very low light levels, we perceive no color, only shades of grey with the rod cells in our eyes. The rod cells are more sensitive to low light levels than the cones. This brightness region is called scotopic vision. In between these two extremes, both rods and cones may be active and this brightness range is called mesopic vision. Here I present data that shows a very dark, moonless night sky away from cities and landscape illuminated by the night sky fall in our mesopic range.

There are 3 types of cone cells, each with a peak response at a different wavelength or color, and the three types are responsible for our color vision. Models of color vision use these three cone types (red, green, and blue) in what is called trichromatic vision or Trichromacy. The rod cells wavelength peak is in between the blue and green cone responses, and recent research shows that a 4-color model may be needed to describe color perception in the mesopic range (e.g. Buck et al., 2000 and references therein)!

It is often said on the internet that the Milky Way is too faint to see color. This idea is easily tested using digital cameras calibrated to measure brightnesses of the sky and landscape at night. I will show that the Milky Way is sufficiently bright to perceive color. Indeed, I and many people I have queried at dark sites observe color in the Milky Way around the galactic core (the brightest part of the Milky Way galaxy) and perceive the color as yellow brown. At latitudes where the Milky Way core (at Declination 25 to 30 degrees south) appears higher in the sky, and from high elevations, the colors are more easily scene (approximately on the equator to 50 degrees south latitude). The Milky Way was named for the color of milk many centuries ago. But milk back then was not the white pasteurized product we buy at the store today--it was typically yellow, not white.

The Magnitude Scale

Astronomers use a brightness scale called magnitudes. It is a logarithmic scale where each unit change in magnitude is a change in brightness of 2.51189 (the fifth root of 100). There are about 1.33 photographic stops per magnitude. More simply, a change in brightness of 100 is 5 magnitudes. The larger the number, the fainter the object. Bright stars in the night sky have brightness around magnitude 0. Young people with good eyesight and a very dark country sky can see stars around magnitude 7.5, or about 1000 times fainter than the bright stars in the night sky. The scale goes negative too for brighter things. The brightest star in the night sky is Sirius at magnitude -1.4. Venus reaches -4.4. The full Moon is about -12.5, and the Sun is -26.7, which is 34.2 magnitudes brighter than the faintest stars generally seen by young observers in a very dark night sky -- a brightness change of 47,000,000,000,000 (47 trillion)!

For extended objects, like the Moon, the Milky Way, a landscape, or a bird in a tree, we perceive surface brightness and contrast changes. Our eyes plus brain are contrast detectors. There is a complex interplay between contrast, the apparent size of an object, and its brightness on both our ability to detect an object and to perceive color.

Astronomers describe surface brightness in terms of stellar magnitude in an angular area, and commonly use magnitudes per square arc-second. If an object is large enough and shows enough contrast with a background, we can perceive surface brightness significantly fainter than 24 magnitudes/square arc-second. The surface brightness where the transition between cone dominant (color, photopic vision) and rod dominant (scotopic vision) is between about 19 to 20 magnitudes/square arc-second, dependent on contrast with the background and apparent size of the object. A very dark night sky overhead has a brightness a little faint than 22 magnitudes/square arc-second in the visual passband (green). Scotopic vision is somewhat fainter than this because the scotopic response lies in between the airglow emission lines so the sky is actually darker then typically reported using photopic (visual) response. The green channel in a digital camera has a response close to the green cone response in our eyes, and is closest to the "visual" passband used by astronomers..

The Bortle scale is a semi-quantitative scale for the darkness of the night sky. The night sky away from cities on a moonless night can reach a little fainter than 22 magnitudes per square arc-second in the visual passband (green). We perceive a Bortle 1 class dark sky as greyish when fully dark adapted, not black. The color of the night sky is also apparent at the magnitude 22 per square arc-second level. For example, I often see the dark country night sky as greenish-gray or reddish-grey depending on the airglow (as in Figure 1) and a few second exposure with a fast lens on a digital camera confirms the color. The night sky typically does not get darker because of airglow by oxygen at 90 to 100 km altitude which emits green light (e.g. commonly seen in aurora) or red light from oxygen emission higher up. A city sky can be 100 times brighter, brighter than 18 magnitudes per square arc-second. Table 1 shows the Bortle scale.

Table 1 The Bortle Scale
Zenith Sky Brightness
(Magnitudes/sq arc-sec)
1 21.7 - 22.0 Black
2 21.5 - 21.7 Grey
3 21.3 - 21.5 Blue
4 20.4 - 21.3 Green
5 19.1 - 20.4 Yellow
6 18.0 - 19.1 Orange
7 18.0 - 19.1 Red
8 <18.0 White
9 <18.0 White
Note: colors assigned vary between implementations.
Colors are used for creating maps of light pollution and
do not represent the color of the night sky.

It should be noted that the color and contrast of an object in the night sky also depends on how high it appears in the sky. Two factors influence this. The atmosphere absorbs more light as the object appears lower in the sky. Blue light is absorbed more than red, so an object becomes yellower and redder as well as fainter nearer the horizon (e.g. the setting sun becomes redder and not as bright as when the sun is high in the sky). Overhead in a clear sky about 15% of the light is absorbed in the green passband by our atmosphere. At 30 degrees above the horizon, 30% is absorbed. The other factor is airglow increases in brightness as one looks close to the horizon, because we are looking through a longer path to space. That increased airglow and light pollution brightness is added light and decreases contrast with objects beyond our atmosphere. Thus, the two factors of absorption by the atmosphere and reduction in contrast from airglow and light pollution as the object appears lower in the sky is destructive to color perception and even simply detecting the object.

The Color and Brightness of the Sky and Landscape at Night

A calibrated digital camera can be used to measure the surface brightnesses of any object in the scene. I have done this calibration and show the digital camera red, green and blue surface brightnesses in a night landscape scene in Figure 1.

Figure 1. A night landscape scene with measurements for the digital camera red, green, and blue color channels are shown in units of magnitudes/square arc-second. For example, in the brightest part of the Milky Way, near the galactic core (right of center in the image), the star clouds indicate a brightness of 19.5 magnitudes/square arc-second in the red channel, 20.0 magnitudes/square arc-second in the green channel, and 20.4 magnitudes/square arc-second in the blue channel. Near the horizon, green airglow is seen and above the green is red airglow. Both these colors are due to emissions from oxygen at two different altitudes, red above about 100 km, green about 90 to 100 km. The landscape shows fainter surface brightnesses, yet the color of the grass was apparent even at a surface brightness fainter than 23 magnitudes/square arc-second. This is a natural color image. Gallery image is here: Summer Milky Way Nightscape with the San Juan Mountains of Colorado.

When the Moon is out, light levels on the landscape and are higher, and color is easier to see. Patio tiles illuminated by a gibbous Moon are shown in Figure 2. The contrast between the tiles and the the grout is low, yet I could easily see the colors of the tiles and the color differences between tiles. I could also see all the colors on the MacBeth color chart. Measurements of the surface brightnesses provides another anchor point in understanding brightnesses where color can be perceived.

Figure 2. A gibbous Moon shines on patio tiles and a MacBeth color chart. Numbers are the red, green and blue surface brightnesses in magnitudes/square arc-second. Note that the brightnesses are only slightly brighter (about 2 magnitudes) than the light from the Milky Way in Figure 1.

At the other end of the brightness range is a clear, dark, moonless night sky away from cities when the summer Milky Way is below the horizon is illustrated in Figure 3. The colors in the chart in Figure 3 were close to what I observed on the chart at the time of the photograph. The darker colors, especially the dark blues appeared gray to me. The green appeared a gray green. The other colors were pastel versions of the colors in the chart. Even the dark brown in the upper left corner was visible as a brownish-gray. Note, the image and the visual observations of the chart were made around midnight on March 22, 2017 from the central Colorado, USA, Rocky Mountains at 10,000 feet elevation, so the brighter summer Milky way was not in the sky, only the fainter winter Milky Way, which was low in the western sky, and mostly blocked by trees. With the brighter summer Milky Way, the colors should be more prominent. Try it yourself. People with normal color vision can see colors at night. I also found that holding the color chart up helped add references, enabling me to distinguish more colors in stars. To see these colors, I used no artificial lights, including camera LCD for over 30 minutes. That allowed my eyes to get dark adapted.

Figure 3. Image of a MacBeth color chart sitting on a car windshield and illuminated only by the night sky (example image of the overhead night sky). Color balance is daylight. Note that the yellow patch was more than 2 magnitudes fainter than light from the Milky Way galactic core region (Figure 1), and the yellow color in the chart could be seen.

Photometry and Threshold Color Perception

Now that I have shown measurements of brightnesses and color for many conditions, how does this compare to published research studies of human perception? The many measurements of surface brightness above are plotted in Figure 4 on threshold contrast detection levels determined in an extensive study by Blackwell (1946), and analyzed by myself in my 1990 book Visual Astronomy of the Deep Sky (see references).

The curves in Figure 4 are lines of apparent size of the object. The kink in the curves around 20 magnitudes per square arc-second is the change where cones dominate (left of the kink) versus rods dominate (right of the kink). But that does not mean one type turns off on the other side of a kink. Cones still operate to the right (fainter) from the kink and rods still operate left (brighter) from the kink. It is this crossover region where both cell types are operating that is mesopic vision. And from the measurements of brightness above, we see that the night sky and the landscape illuminated by the night sky fall in the mesopic range.

How to determine detectability of an object? For example, the core of the Milky Way appears several degrees across. The brightness of the core was measured and the brightness of dark regions around the core were measured to determine the contrast difference. In a dark country sky, the Milky Way core has a log contrast around 0.5 and a surface brightness of 20.0 magnitudes/square arc-second (see Figure 1) in green light and an adjacent background of about 20.9 magnitudes/square arc-second. The contrast threshold at that background surface brightness is between 8 and 16 arc-minutes (less than 1/4 degree). Because the Milky Way core is larger than that, it is well above the threshold and color can be seen. The 1946 Blackwell data agree with modern observations.

Figure 4. Contrast thresholds of the human eye derived from data in Blackwell (1946). Contrast is defined as object surface brightness, B, minus background surface brightness, Bo, divided by the background surface brightness. The curves are for different apparent sizes of the object. If contrast is above a line, the object is detected. In the photopic range full color is perceived. In the mesopic range colors appear more pastel as brightness decreases, and color perception disappears in the scotopic range. Note that the Milky Way brightness puts it in the mesopic range, so color should be seen. The Bortle scale is a semi-quantitative scale for the darkness of the night sky. Note too that the Milky Way actually appears brighter in a light polluted sky (B9) than in a dark country sky (B4) because of the added light pollution, but the contrast is low in the B9 skies, so color is not perceived. A complicating factor is also distinguishing color next to similar color. The yellowish Milky Way against an orange light pollution sky is harder to perceive at low contrast because the colors are close.

The threshold detection data in Figure 4 were a significant part of my 1990 book Visual Astronomy of the Deep Sky, and the interpretation for detecting objects in a telescope at different magnifications. Extended objects like galaxies and nebulae do not have a higher surface brightness in a telescope--the surface brightness is always lower. The contrast does not change either, and it does not change with increasing magnification. But the magnification makes things appear larger, thus moving horizontally to the right in the Figure 4 plot, and that makes detection of a larger low contrast object easier. The key is for a given contrast, to make the object larger. The 1990 book is out of print, but for more on this subject, see: Optimum Magnified Visual Angle, Visual Astronomy of the Deep Sky.

Discussion and Conclusions

From the data presented above, we see that the main difficulty in perceiving color in the night sky is low contrast, not low brightness. The darker the skies, the better the contrast so colors may be seen easier. The amount of airglow varies from night to night, and with location on the Earth, so cherish those very clear moonless nights away from the light pollution of cities and with low humidity and low airglow. While one can travel to dark high altitude low humidity locations, airglow occurs at altitudes much higher than any mountain so one can not escape airglow effects unless you travel in a spaceship above 300 kilometers.

Airglow occurs all over the Earth. At equatorial latitudes, red airglow dominates from oxygen emission 100 to 300 km high. At high latitudes north and south, green airglow dominates from oxygen emission 90 to 100 km high (300,000 to 330,000 feet altitude). In between (mid latitudes), both red and green airglow is commonly seen, as in Figure 1. Airglow occurs all year, all night, and at solar maximum and solar minimum.

Colors in a dark, moonless night sky away from cities and in the landscape illuminated by a moonless night sky can be perceived. Key to perception in low light conditions is to dark adapt your eyes for at least 30 minutes (45 minute is better) with no other light, including flashlights, or lights from camera LCDs. When colored artificial lights are used, they warp color perception. For example, use of a red light makes other things appear bluer. For proper color perception, use neutral color light or at least not strong colors (like red). I use very dim yellowish lights when I need a light. Turn down the brightness on smartphones, computer screens and camera LCDs when using them at night to help maintain your night vision. Of course use lights as needed for safety, including walking on a trail. Once dark adapted, any light will impact that adaption, and the brighter the light, the longer it will take to get dark adaption back. A few seconds of light can mean 5 to 10 minutes are needed to return dark adaption. For more information, see Lighting and Protecting Your Night Vision.

With good dark adaption at a dark site away from cities on a moonless night an amazing number of stars in the sky can be seen, and intricate details of the Milky Way when it is up. And colors in the Milky way will be seen (pastel colors). Stars too will show color. Star colors can be seen easier with binoculars or a telescope with the eyepiece slightly out of focus so that the stars appear as little disks. Binoculars and telescopes can show even more colors.

If you find the information on this site useful, please support Clarkvision and make a donation (link below).

References and Further Reading

Buck, S. L., R. F. Knight, and J. Bechtold, 2000, Opponent-color models and the influence of rod signals on the loci of unique hues, Vision Research 40 p3333-3344.

Blackwell, R.H., Contrast Thresholds of the Human Eye, Journal of the Optical Society of America, 36, p624-643, 1946.

Clark, R.N., Visual Astronomy of the Deep Sky, Cambridge University Press and Sky Publishing, 355pp., 1990. Astrophoto Gallery. Nightscapes Gallery.

The Night Photography Series:

Home Galleries Articles Reviews Best Gear Science New About Contact

First Published December 28, 2018
Last updated December 28, 2018