If you took photos on Mars, the light would be about two stops dimmer than a sunny day on Earth, because Mars receives only 43% of Earth's sunlight and a dusty sky absorbs more. The sky would look butterscotch by day and blue around the setting Sun, and your camera's automatic white balance would get the colours wrong. NASA's rovers solve this with a calibration target, a small chart of known colours and greys on the rover deck that every colour photo is checked against.
- 586.2 W/m²: sunlight at Mars's average distance, against 1,361 W/m² at Earth (NASA). That alone costs 1.2 stops.
- About 1/26 s at f/16, ISO 100: my "sunny 16" for Mars at noon with a typical dust optical depth of 0.5.
- Tau 10.8: the dust opacity Opportunity reported in June 2018, the storm that ended its mission.
- July 1976: Viking 1's first colour picture showed a blue sky because the team had to calibrate by guesswork within eight hours.
- 8 patches over 8 magnets: how Perseverance's calibration target keeps its reference colours free of dust.
Mars photography is a real job. Perseverance's Mastcam-Z has been shooting since February 2021, and NASA released a 46-image panorama from a site called "Arbot" in May 2026. But a human with a camera on Mars would meet three problems at once: less light, a coloured sky, and dust. Let me work through each with the numbers, then design the kit and settings that would work.
How much light is there on Mars?
Less than half of Earth's, before the dust. NASA's Mars fact sheet gives a solar irradiance of 586.2 W/m² at Mars's mean distance of 227.956 million km from the Sun, against 1,361.0 W/m² at Earth (NASA's Earth fact sheet): a ratio of 0.431.
Calculated in this post from the two NASA fact-sheet irradiances: 586.2 ÷ 1,361.0 = 0.431.
That ratio follows from the inverse-square law: light from a point source spreads over a sphere, so intensity falls with the square of distance. Mars orbits 1.524 times farther out than Earth, and \( 1/1.524^2 = 0.431 \).
Photographers measure light in stops, each a factor of two. Losing 1.22 stops is like going from a bright afternoon into light shade. A human eye would barely notice it, because the iris and retina adapt. A camera on manual settings would.
| Condition | Earth | Mars | What it does to a photo |
|---|---|---|---|
| Sunlight at top of atmosphere | 1,361 W/m² | 586.2 W/m² | 1.2 stops darker |
| Surface pressure | 1,013 mb | 6.36 mb | Thin air, but full of fine dust |
| Main gas | Nitrogen | 95.1% CO₂ | Little Rayleigh blue from gas |
| Average temperature | Far milder | About −59 °C (214 K) | Batteries and screens suffer |
| Daytime sky | Blue | Butterscotch | Auto white balance fails |
| Sunset near the Sun | Orange-red | Blue | Reversed colour cues |
Why are Mars photos so hard to colour correctly?
Because the light itself is tinted, and a camera cannot tell tinted light from a tinted object. On Earth, your camera assumes something in the scene averages to grey under bluish-white daylight. On Mars, the sky scatters reddish dust light onto everything, so the "grey" assumption pushes the whole image wrong.
On 21 July 1976, a day after Viking 1 landed, the orbiter relayed the first colour picture ever taken from the surface of Mars. The imaging team had to release prints within eight hours. They used the same simple voltage normalisation that had worked in pre-flight tests, and the result was startling: the ground and the sky were both bathed in a reddish glow. Unwilling to publish something that looked like an error, they adjusted the calibration until the sky came out neutral grey, and in reproduction it turned bluish. That Earth-like blue sky went round the world. Further analysis showed the reddish cast was real, from fine soil particles suspended in the air, and several days later a second version with a reddish sky was released. Newspapers ran headlines about the Martian sky changing from blue to red.
NASA SP-425, The Martian Landscape: The first color picture
The fix: carry a colour chart
Every NASA rover camera since then has carried a calibration target. Mastcam-Z's primary target, a gold-plated aluminium base 98 mm across, holds eight ceramic colour and grey patches set over eight samarium-cobalt magnets, which repel the magnetic Martian dust from their centres. Four grey rings surround a central shadow post, or gnomon, painted black.
Describes the two deck-mounted targets that let Mastcam-Z convert raw image data to reflectance on a tactical timescale. The primary target (45.5 mm tall) has eight ceramic patches over Sm₂Co₁₇ magnets, four grey rings and a gnomon whose shadow separates direct sunlight from diffuse sky light; a secondary angled shelf carries seven vertical and seven horizontal chips as a cross-check. The authors note that on Curiosity's landing, the descent engines deposited dust equal to an optical depth of a little over 0.1, about 25 sols of dust accumulation in moderately dusty conditions, and that the target's area became one of the dustiest on the deck. The limit: the dust correction relies on a simplified scattering model whose parameters are proxies, not true dust properties.
Read the paper (free)Worked example: turning a raw image into true colour
The method is a version of von Kries white balancing: scale each colour channel so that a known reference comes out right. Because the target's reflectance \( R_{cal} \) is measured on Earth, its brightness in an image tells you the illumination in each channel.
- Measure the light. For one channel, the camera signal from a target patch is \( DN_{cal} = k \cdot J \cdot R_{cal} \), where \( J \) is the irradiance and \( k \) the camera's response.
- Solve for it. \( kJ = DN_{cal} / R_{cal} \). For a patch of known reflectance 0.40 reading 1,200 counts, \( kJ = 3{,}000 \).
- Apply to the scene. A rock reading 600 counts in the same channel has reflectance \( R = 600 / 3{,}000 = 0.20 \).
- Repeat for every filter. Doing this in red, green and blue (and Mastcam-Z's narrow bands) gives colours independent of the dusty light.
The numbers in steps 2 and 3 are illustrative. The real pipeline also uses the gnomon's shadow to split direct from diffuse light and corrects for dust on the target. The result is a choice of renderings: natural colour, as a person on Mars would see it, or white-balanced, as the same rocks would look under Earth daylight, which helps geologists compare with samples at home.
How much does Martian dust darken a photo?
From a little to almost totally. Dust opacity is measured as optical depth, tau (\( \tau \)), and the Beer–Lambert law says the direct sunbeam is weakened by a factor of \( e^{-\tau m} \), where \( m \) is the air mass, about \( 1/\sin(\text{sun elevation}) \).
Spirit and Opportunity photographed the Sun through narrow filters almost every sol, and fitting the Beer–Lambert law to those images gave a five-Mars-year record of optical depth. Optical depths were low and steady around aphelion (Spirit fell from about 0.9 to below about 0.3 early in the mission; Opportunity to below about 0.5) and high and variable near perihelion. The 2007 global storm drove optical depth to 4.3 at Spirit, with tau around 3 or more for 30 sols, and to a peak of 4.6 at Opportunity on 16 July 2007, after which the rover hibernated. Dust particles had effective radii of about 1 µm in northern summer and about 2 µm when dust was being lifted. The limit: two sites only, and the Sun can't be imaged at the very highest opacities.
Read the paper (free)See the data as a table
| Optical depth (tau) | |
|---|---|
| Spirit, clear aphelion season 2004 | 0.3 |
| Opportunity, clear aphelion season 2004 | 0.5 |
| Spirit, 2007 global storm peak | 4.3 |
| Opportunity, 2007 global storm peak | 4.6 |
| Opportunity, June 2018 storm | 10.8 |
Source: Lemmon et al., Icarus 2015 (2004 and 2007 values); NASA JPL (2018) · 2004 values are the approximate lower bounds reported. Direct sunlight falls as e^(-tau): tau 0.5 passes 61%, tau 4.6 passes 1%.
Worked example: Mars's "sunny 16"
On Earth the sunny 16 rule sets f/16 and a shutter of 1/ISO in bright sun. To adapt it, add the distance loss to the dust loss, both in stops, and treat Earth's own clear-air loss as already built into the rule.
- Noon, typical dust. \( \tau = 0.5,\ m = 1 \): \( 1.22 + 0.72 = 1.94 \) stops. At ISO 100, \( t = 2^{1.94}/100 \approx 1/26 \) s.
- Mid-afternoon. Sun at 30°, \( m = 2 \): 2.66 stops, about 1/16 s.
- Near sunset. Sun at 10°, \( m = 5.76 \): 5.37 stops, about 0.4 s. A tripod job.
- 2007-style storm. \( \tau = 4.6 \): 7.85 stops, about 2.3 s.
Sensitivity check: where the model breaks
Plug in Opportunity's tau of 10.8 and the formula gives 16.8 stops, a 19-minute exposure. That is wrong in practice, because the formula counts only the direct beam. In thick dust most light arrives scattered, as diffuse sky glow, which is exactly why Mastcam-Z's target has a gnomon to measure the two parts separately. Above a tau of about 1, the direct-beam rule overstates the darkness; meter the scene instead.
Use this to check what you just read: set the dust, the Sun's height and your camera, and get a starting exposure for Mars.
Exposure on Mars (direct sunlight)
Why is a Martian sunset blue?
Because Martian dust grains are about the size of the light's wavelength, and particles that size send blue light forward, close to the Sun's direction. The fine dust lets blue light through the atmosphere more efficiently than red, so the sky right around the setting Sun glows blue while the rest of the sky stays yellow-brown.
This is Mie scattering, the regime for particles comparable to the wavelength (Lemmon's dust radii of 1 to 2 µm against 0.4 to 0.7 µm light), and it behaves differently from the Rayleigh scattering by tiny gas molecules that makes Earth's sky blue. Mars's thin CO₂ adds little Rayleigh blue, so the dust controls the colour. The effect is strongest at sunset because the light path through the dust is longest.

Curiosity recorded its first colour sunset on 15 April 2015, sol 956, with the left Mastcam. NASA noted the camera is slightly less sensitive to blue than a human eye, so a person standing there would see the blue a little stronger than the photo shows.

What camera would work on Mars?
One built like Mastcam-Z: modest resolution, superb calibration, sealed against dust. Each Mastcam-Z camera uses a 1,648 × 1,214 pixel CCD, about 2 megapixels, with a 26–110 mm zoom at f/6.7 to f/9.5. That is less than any phone, yet it resolves 0.7 mm features at 2 m because the optics and calibration are excellent.
Calculated in this post from the 1,648 × 1,214 pixel count in Bell et al. (2021): 1,648 × 1,214 ≈ 2.0 million pixels.
Describes the twin zoom cameras on Perseverance's mast, 24.4 cm apart for stereo. Each has zoom, focus and filter-wheel mechanisms and a 1,648 × 1,214 CCD; fields of view run from 25.6° × 19.2° at 26 mm to 6.2° × 4.6° at 110 mm, resolving 0.7 mm at 2 m and 3.3 cm at 100 m. Besides Bayer RGB colour, narrow filters sample up to 11 wavelengths from 400 to 1,000 nm to map iron minerals, and neutral-density filters (attenuating by 10⁵ and 10⁶) let it photograph the Sun directly to measure dust optical depth. A requirement: signal-to-noise of at least 30:1 at a 0.015 s exposure on dark targets with the Sun low. The limit: a 2 MP sensor trades resolution for heritage and reliability.
Read the paper (free)Your Mars camera bag
- A grey card or colour chart in the first frame of every scene, since the sky colour changes with dust.
- Shoot RAW, so white balance is a choice made later from the chart, not a camera guess (see RAW and dynamic range).
- A sealed body and no lens changes outside: Kinch's team found landing alone dusted the deck to an optical depth of about 0.1.
- Warm batteries: the average surface temperature is about −59 °C, and even in the 2018 storm Opportunity sat at about −29 °C.
- A tripod for the golden hour: near sunset the light can be 5 stops below Earth's sunny 16.
- ND filters for the Sun itself: Mastcam-Z uses 10⁵ to 10⁶ attenuation to image the solar disc.

On Mars, the sky is in every photo even when it is out of frame, because it colours all the light.From this post
Ordinary photography works on Mars, but not on Earth settings. Sunlight is 43% of Earth's (1.22 stops) and typical dust at tau 0.5 takes another 0.72 stops at noon, so a Martian "sunny 16" is about 1/26 s at f/16 and ISO 100, sliding to about 0.4 s near sunset. Auto white balance will fail under the dust-tinted sky, as Viking's team learned in 1976, so the essential change is a calibration reference in frame, exactly what Mastcam-Z's eight-patch target provides, plus a sealed, warmed camera body.
Calculated in this post from NASA's irradiances (Mars, Earth) and an example dust optical depth of 0.5 with the Sun overhead (with these example inputs, the worked example above gives 1.22 + 0.72 = 1.94 stops, so 21.94 ÷ 100 ≈ 1/26 s).
What would your photos look like back on Earth?
Strange, unless you decide what "true" means. A natural-colour Mars photo looks dim and brownish to Earth eyes; a white-balanced one looks like a desert at home. Neither is wrong; they answer different questions. The same choice exists in every photo you take, as I cover in the exposure triangle in stops and how phone cameras compute a photo, where software makes the colour decision for you.
If you want to practise the Martian workflow, shoot a dusty Indian summer evening, when haze turns the sky beige, with a grey card and a tripod. For more on photographing on the road, see my motorcycle travel photography guide and the Kala world page.
What's new
- Perseverance's Mastcam-Z stitches 46 images into a panorama at "Arbot", beyond Jezero Crater NASA, May 2026
- A public-processed Mastcam-Z panorama from sol 1771 shows how anyone can work with raw Mars images ASU Mastcam-Z gallery, Feb 2026
- Curiosity's Sun-imaging record of dust optical depth in Gale Crater, with sand reaching 1.97 m arXiv (Lemmon et al.), Sep 2023
What if next?
- What if humans could see ultraviolet and infrared? Mastcam-Z already sees to 1,000 nm. What would our own eyes need?
- What if every camera on Earth vanished? The rovers would be the last working cameras in the solar system.
- What if you had to design a logo that lasts 10,000 years? Another kind of calibration problem: making a sign mean the same thing far away in time.
Questions people ask
What colour is the sky on Mars?
By day it is a butterscotch or tan colour, from suspended dust. Near sunrise and sunset, the area around the Sun turns blue, because the fine dust scatters blue light forward.
Are NASA's Mars photos true colour?
NASA releases both natural colour, approximating what a person would see, and white-balanced or enhanced versions for science. Each colour image is calibrated against the rover's onboard target of known colours.
Would a normal camera work on Mars?
Optically yes, but cold (about −59 °C on average) and dust would shorten its life, and auto white balance would misread the tinted light. A sealed, heated body and a grey card would help.
Is Mars darker than Earth?
Yes. It receives 43% of Earth's sunlight, and dust removes more, so a typical clear noon is roughly two stops darker than a sunny day on Earth.
Sources
- NASA NSSDC: Mars fact sheet (586.2 W/m², 1.524 ratio, 6.36 mb, 214 K, 95.1% CO₂)
- Kinch et al., Space Science Reviews: Radiometric calibration targets for Mastcam-Z (2020) (target design, landing dust)
- Bell et al., Space Science Reviews: Mastcam-Z investigation (2021) (sensor, zoom, f-numbers, filters)
- Lemmon et al., Icarus: Dust aerosol, clouds and optical depth over 5 Mars years (2015) (optical depths, 2007 storm, particle sizes)
- NASA SP-425, The Martian Landscape: The first color picture (Viking 1 colour story)
- NASA JPL: Opportunity hunkers down during dust storm (2018) (tau 10.8, −29 °C)
- NASA Photojournal: Sunset in Mars' Gale Crater (2015) (blue sunset explanation, Mastcam blue sensitivity)
- Wikipedia: Sunny 16 rule (Earth exposure baseline)
- NASA: Perseverance captures panorama at Arbot (2026)

