00THE PHYSICS OF GRAIN

Why dim photos
look grainy.

Raindrops falling on pale stone paving, rippling a shallow puddle on one side and scattering beads of water at random across the dry slabs — AI-generated image, not a photograph of a real person or place.

Light arrives in random drops.

In bright light the randomness evens out. In dim light each pixel catches only a few photons, and the count jumps about from pixel to pixel: that is the speckle you see. Physics courses and photography tutorials explain why, and what helps.

√Nthe square-root ruleUpdated 18 October 2026

Check a shot

01Principles

More light, less speckle
Lowest ISO you can
Average several frames
Random noise, tamed by numbers

02THE SHORT ANSWER

Too few photons,
too much chance.

Grain is mostly statistics.
Three ideas explain most of it.

SHOT NOISE

Light arrives in drops

Photons arrive discretely, independently and randomly, like the first drops of a shower. Catch only a few and the count varies from pixel to pixel: that variation is noise.

See the numbers

ISO

Turning up ISO amplifies

A higher ISO works by amplifying the signal from the sensor, and that amplifies the noise too. Higher ISO speeds give progressively more noise.

Check a shot

AVERAGING

Stack frames to smooth it

Random noise rises and falls around the true value, so averaging identical shots evens it out. Four frames halve it, a trick astrophotographers rely on.

How to do it

03A QUICK SHOT CHECK

What kind of
shot is it?

Pick a situation to see which kind of noise to expect. The bar shows how much visible noise the tutorials say it tends to produce.

Based on Cambridge in Colour's tutorials on digital camera image noise and image averaging. General photography information.

Shot check

Choose a situationBright daylight, low ISOA dim room at high ISOA long exposure on a warm nightBrightening dark shadows laterFour identical frames, averaged

Bright daylight, low ISOBarely visible

Plenty of light gives a high signal-to-noise ratio. With a high ratio the noise is so small next to the picture that it can seem to be missing.

A dim room at high ISORandom speckle

Random noise is always present and is most influenced by ISO speed. Its pattern changes from frame to frame, even with identical settings.

A long exposure on a warm nightHot pixels

Very long exposures can show fixed pattern noise: bright 'hot pixels' that heat makes worse. They sit in the same places each time, so they are the easiest to subtract.

Brightening dark shadows laterBanding

Banding noise comes from reading data off the sensor. It shows most in shadows, at high ISO, or when an image has been brightened a lot.

Four identical frames, averagedHalf the noise

Random noise falls by the square root of the number of frames averaged, so four frames cut it in half while keeping fine detail.

Source: Cambridge in Colour, Digital camera image noise (part 1) and Noise reduction by image averaging.

04TWO NUMBERS WORTH KNOWING

Noise follows
a square root.

One number on counting photons,
one on averaging frames.

≈ 3.2 electrons

Noise on 10 photo-electrons

Photon arrivals follow Poisson statistics, so the noise is the square root of the average count. A star giving 10 photo-electrons a second has noise of about 3.2 in one second; over 10 seconds it gives 100, with noise of 10.

PHOTON NOISE · RIT DETECTORS COURSE

4 frames

Frames averaged to halve random noise

Noise fluctuation drops by the square root of the number of images averaged. Two averaged frames at ISO 400 are comparable to one taken at ISO 200.

AVERAGING · CAMBRIDGE IN COLOUR

The CCD image sensor was invented in 1969 by Willard Boyle and George Smith at Bell Laboratories, and earned them a share of the 2009 Nobel Prize in Physics.

05FOUR HABITS TO TRY

Cleaner pictures,
same camera.

None of these needs new equipment. They follow from the square-root rule: more light and more frames beat amplifying a weak signal.

Collect more light
When photon noise dominates, the signal-to-noise ratio grows with the square root of the exposure time, so a longer exposure helps.
Use the lowest ISO you can
Higher ISO amplifies the noise along with the picture, so keep it as low as the scene allows.
Average several frames
On a tripod, take identical shots and average them. It raises the signal-to-noise ratio without softening detail; it does not suit hand-held shots.
Expect more hot pixels when warm
Heat makes fixed pattern noise worse in very long exposures, and cooling a sensor reduces dark current.

A dim room at night lit by a single candle on a wooden table, with deep shadows all around — AI-generated image, not a photograph of a real person or place.

06THE OTHER NOISES

Not all noise
comes from light.

A clear night sky crowded with stars above a dark hillside, with the silhouette of a tripod in the foreground — AI-generated image, not a photograph of a real person or place.

A sensor is not silent even in the dark. Read noise comes from the electronics as each pixel's charge is read out, and it does not depend on how much light arrived. Dark current is charge made by heat, and it builds up during long exposures.

Lund University's lab notes say photon shot noise is the main noise source, that dark current noise is not really relevant until exposures of several seconds, and that fixed pattern noise can now generally be ignored in consumer cameras.

07Questions

Questions people ask

Is digital noise the same as film grain?

Cambridge in Colour calls image noise the digital equivalent of film grain. It can spoil detail, but some photographers like it for an old-fashioned, grainy look.

Why do photos get grainy in dim light?

Photons arrive randomly, and the noise is the square root of how many a pixel collects. With few photons that randomness is large compared with the picture, so it shows as speckle.

Does a higher ISO cause the noise?

A higher ISO amplifies the image signal in the camera, and that also amplifies the noise, so higher ISO speeds produce progressively more noise.

What are the bright dots in long night exposures?

Often hot pixels, a kind of fixed pattern noise that shows in very long exposures and gets worse with heat. Because they repeat in the same places, they can be subtracted.

Who first described shot noise?

Walter Schottky introduced the idea in 1918 while studying fluctuations of current in vacuum tubes. The same statistics apply to photons reaching a sensor.

08Sources

Why dim photos
look grainy.