Dynamic Range Explained

Dynamic range is the span between the darkest and the brightest tones that can be recorded with detail, measured in stops. It matters because when a scene has more range than your camera, something has to give: the brightest parts clip to pure white, the darkest sink to black, or both, and every decision about exposure in strong light comes down to which.

Every stop is a doubling of light, so the numbers grow fast. A camera that holds 12 stops can keep detail in highlights about 4,096 times brighter than its darkest usable shadows. A sunset with the sun in the frame, like the one above, can run past 20 stops. No single exposure holds all of that, so the photographer chooses what to keep.

Three dynamic ranges: the scene, the camera and the output

The term is used in three places, and most confusion comes from mixing them up.

• The scene has a dynamic range: the ratio between its brightest and darkest parts. A sunlit sky above a shaded valley has a lot, which is what people mean by a high dynamic range scene. A foggy morning has very little, a low dynamic range scene that fits in one exposure with room to spare.
• The camera has a dynamic range: how much of that ratio the sensor can record in one exposure, from the point where it clips to the point where detail disappears into noise.
• The output, a screen or a print, has its own range, usually smaller than the camera’s. A glossy print reflects roughly 100 to 200 times more light from its whitest paper than from its blackest ink, about 7 stops. A matte print manages closer to 5 or 6. A typical screen in a lit room shows around 8 to 10.

Most of the work in high contrast light is fitting the first into the second, then the second into the third. That last squeeze is why a raw file that holds 12 stops still has to be compressed in editing before it looks right on a screen.

Dynamic range is also not the same as contrast. Contrast describes how the tones in a picture are spread; a high contrast photo can come from a low range scene with the tones pushed apart in editing. Tonal range describes which tones a finished image actually uses. Dynamic range is the size of the container they all have to fit in.

See it: fit a scene into the sensor

The tool below lays the sensor’s window over a scene measured in stops; slide the exposure and watch what clips at each end, change how many stops the sensor holds, then bracket three frames and merge them.

Dynamic range in stops: what the numbers mean

Because each stop doubles the light, a difference of a few stops is a large difference in what a camera or a scene holds. The middle column is 2 multiplied by itself that many times.

• 5 stops: Brightest to darkest: 32 to 1. A scene with about this range: Fog, or a flat overcast scene with no sky in the frame.
• 7 stops: Brightest to darkest: 128 to 1. A scene with about this range: Soft window light on a subject, no window in the frame.
• 10 stops: Brightest to darkest: 1,024 to 1. A scene with about this range: A sunlit landscape with some shade.
• 12 stops: Brightest to darkest: 4,096 to 1. A scene with about this range: A backlit portrait, or a sunset over land with the sun out of frame.
• 14 stops: Brightest to darkest: 16,384 to 1. A scene with about this range: Deep forest shade with sunlit sky showing between the trees.
• 17 stops: Brightest to darkest: 131,072 to 1. A scene with about this range: A dim room with a sunlit window in the frame.
• 20 or more stops: Brightest to darkest: 1,048,576 to 1. A scene with about this range: Streetlights and deep shadow at night, or the sun itself in the frame.

The scene values are rough guides; real scenes vary a lot. The shape is what counts. Everyday scenes run from about 5 to 20 stops or more, and a single raw exposure from a good sensor at base ISO covers somewhere in the middle of that, roughly 10 to 14 stops depending on how it is measured.

What sets a camera’s dynamic range

Think of each photosite on the sensor as a bucket catching light. The top of the bucket is the ceiling: once it is full, it records pure white and any more light is lost, which is clipping. The bottom is a floor of electronic noise that the sensor and its circuits add whether or not light arrives. Detail darker than that noise cannot be told apart from it. Dynamic range is the distance between the two, counted in stops.

With real numbers: a sensor whose photosites hold about 65,700 electrons before filling, with a read noise of about 2.5 electrons, has a ratio of 26,280 to 1. That is about 14.7 stops, because 2 multiplied by itself 14.7 times is 26,280. Everything below changes either the ceiling or the floor.

• ISO. A camera has its widest range at its lowest native ISO. Raising ISO amplifies the signal, so highlights in the scene clip sooner. On some sensors the shadow noise barely changes as ISO rises, and each stop of ISO costs close to a full stop of range. Such sensors are called ISO invariant, and on them you can keep the ISO low to protect a bright sky, then brighten the dark parts in editing with about the same noise you would have got by raising the ISO in the camera. On many others the shadow noise also drops at higher ISO, so the loss is smaller at first, sometimes only a stop or so across the first four stops of ISO. Sensors with dual native ISO have two gain settings and switch to the second at a set ISO that varies by camera; just above the switch the shadow noise drops, so the range there can be wider than just below it. Either way, base ISO holds the most.
• Sensor size and design. A larger sensor collects more light in total. At the same technology and the same print size, a sensor with about 2.25 times the area (the step between the two most common interchangeable lens formats) gains roughly one stop, and four times the area gains roughly two. Design matters at least as much, which is why a newer small sensor can match an older large one. See sensor size for the wider trade-offs.
• Bit depth. The converter that turns the signal into numbers sets a cap. A 12 bit file has 4,096 levels and cannot record more than about 12 stops; a 14 bit file has room for about 14. More bits do not create range the sensor did not capture, but too few can throw some away. Bit depth covers how levels are spread across the tones.
• Raw or JPEG. A raw file keeps everything the sensor recorded. A camera JPEG commits to one contrast curve and clips whatever falls outside it, so it keeps noticeably less recoverable highlight and shadow detail than the raw from the same shot. If a scene is close to the camera’s limit, shoot raw.

Two common claims are wrong. More megapixels do not give more dynamic range, and an expensive lens does not add any either, although a lens that flares lowers contrast and fogs the shadows.

Why published numbers disagree

You will see the same camera listed at 12 stops in one place and 15 in another. Both can be honest. An engineering measure counts down to the point where the signal equals the noise (a signal-to-noise ratio, or SNR, of 1), which is the floor of the bucket and looks awful in a photo. That is the link between the two terms: SNR describes how clean one tone is, and dynamic range is how many stops lie between clipping and the tone where SNR falls to the chosen limit. A photographic measure stops higher, where shadows are still clean enough to print, and usually scales to the same print size so cameras with different pixel counts compare fairly. The same sensor can measure about 12 stops one way and about 9 the other. Compare cameras only on the same test, and treat any single brochure figure as a ceiling you will not reach in practice.

Why does a lifted shadow look grainy while a bright area looks clean? The ISO guide explains where the noise in a dark exposure comes from and when brightening it later costs nothing.

Your eye, your camera and your phone

The most common complaint about dynamic range is “it looked fine when I was standing there.” It did, because your eye works differently from a sensor. At any one moment, an eye adapted to the light holds roughly 10 to 14 stops, about what a good sensor holds. The much larger figures you see quoted, 20 stops and more, describe the eye adapting over seconds and minutes. Your eye also never takes one frozen picture: it jumps between the bright sky and the dark ground, adjusting each time, and your brain stitches the views together. The camera makes one exposure with one setting for everything. The eye versus camera page goes further into the differences.

Phones close the gap a different way. Their sensors are small and hold less range per frame, so the camera quietly takes a burst of frames at different exposures and merges them before you see the picture. That is computational photography, and it is why a phone can sometimes hold a bright sky that a larger camera clips in a single frame. The merged result is still squeezed to fit the screen, which is why phone photos of bright scenes can look flat.

How to tell if a scene is beyond your camera

Do not judge by the screen on the back of the camera; its brightness depends on its setting and the light around you. Use the tools made for it.

• The histogram. If the histogram is pressed against both edges at once, with a spike at the right (the highlights) and a spike at the left (the shadows), the scene has more range than that exposure holds.
• Highlight warnings. Turn on highlight warnings so clipped areas flash in playback. A blinking sky tells you exactly what was lost. The reverse is useful too: if nothing blinks, even in the brightest part that matters, you have room to give more exposure and get cleaner shadows.
• Spot readings. Spot meter the brightest area that must keep detail and the darkest area that must keep detail. The difference in stops is the scene’s working range.

The histogram and warnings are built from a JPEG preview, so a raw file usually has a little more highlight room than they suggest. Learn how much your own camera has by testing it, not by guessing. The same kind of test shows how far its shadows stretch: shoot one scene correctly exposed and then 1, 2, 3, 4 and 5 stops darker, brighten each dark frame back in editing, and compare them at 100% to see where the noise stops being acceptable to you.

Practise reading the graph on real photos in the histogram game below, and notice which ones are pressed against an edge.

A stop map of a room with a window

Here is what spot readings of a typical sunny day indoors might look like, written as exposure values at ISO 100. Each step of 1 EV is one stop. These are example figures to show the method; your own room will differ.

The clouds and the shadow under the desk are 11 stops apart, a ratio of 2,048 to 1 between two readings. As a rough guide: if the two areas you care about are within about 5 stops, one frame holds both easily, even as a JPEG. Between about 5 and 9, a raw file exposed to protect the highlights holds both, with the shadows lifted in editing and some noise to show for it. Beyond that, plan one of the methods further down. In this room, the lawn and the far wall (6 stops apart) fit in one raw frame; the clouds and the desk do not.

Exposing for the most range in one shot

Highlights and shadows fail differently. A clipped highlight is gone; there is no data to recover. A dark shadow still has data, just noisier. That is why the usual advice is to protect the highlights and let the shadows fall where they fall.

1. Set base ISO.
2. Give as much exposure as you can without clipping the highlights that matter. This is called exposing to the right, because it pushes the histogram toward the right edge. It means more light reaching the sensor, not a brighter looking picture: you may darken the result again in editing, and the shadows come out cleaner for it.
3. Decide what is allowed to clip. A bare light bulb or the sun itself can go white. A white dress or a cloud should not.
4. In editing, pull the highlights down and lift the shadows. Recovering an overexposed highlight works only where at least one color channel was not clipped; where all three are full, there is nothing left to bring back.

Lifting the shadows of one raw file is not the same as HDR. It moves the tones you already captured into a range a screen can show, but it cannot add range the frame did not record.

Why exposing to the right gives cleaner shadows

Light itself arrives unevenly. Count the photons landing on a photosite and the number jitters from one moment to the next, and that jitter grows more slowly than the light does: four times the light gives only twice the jitter. So more exposure means a cleaner signal in every tone, and the darkest tones gain most because they start nearest the noise floor. When you expose to the right and then darken the picture in editing, the shadows keep the cleaner signal they were captured with. A frame exposed “correctly” by the meter in a high contrast scene throws some of that away.

You will also read that a sensor gives half its raw levels to the brightest stop, a quarter to the next, and so on down. That is true of a linear raw file, and it is why the darkest stops look coarse when they are pushed hard, but the noise in the light is the bigger reason the shadows suffer.

Three conditions apply. Exposing to the right means more light, from a slower shutter speed or a wider aperture, not a higher ISO, which only amplifies what you already caught. It needs a raw file, because a JPEG cannot be darkened again without losing quality. And it matters most when the scene is close to the camera’s limit or the shadows will be lifted; in soft, low range light a normal exposure already sits comfortably inside the sensor.

Wondering how to make the exposure brighter without leaving your shooting mode? Exposure compensation explains what the plus and minus control changes in each mode.

Want the detail on reading the graph? The histogram guide shows what each shape means and how to read it in the field.

When the scene has more range than the sensor

Sometimes no single exposure will hold the bright sky and the dark foreground together. You have five ways to deal with it, and the first is often the best.

1. Choose what to lose

Not every picture needs detail everywhere. Expose for the sky and let a figure or a rock become a silhouette, or expose for a face and let the window behind it go white. A clear decision often makes a stronger picture than a compromise that leaves everything a little too dark and a little too bright. Backlighting and silhouettes covers doing this on purpose.

Test your eye on that decision: pick the photo below that metered for the subject rather than the background, then read why.

Which photo metered for the subject rather than the background?

A B

2. Darken the bright part with a graduated filter

A graduated ND filter is dark at the top and clear at the bottom. Lined up on the horizon, it holds back a bright sky by two or three stops while the land is untouched, so one exposure fits. A plain ND filter does not help: it darkens the whole frame equally and leaves the range unchanged.

3. Bracket and merge

Bracketing means taking the same frame at several exposures, then merging them into one image that holds the range of all of them. A merge of frames spread across N stops adds about N stops to what one frame holds. Three frames at minus 2, 0 and plus 2 span 4 stops, so a sensor that holds 12 stops in one frame can cover a scene of about 16. Use a tripod, keep the aperture the same across frames, and change only the shutter speed. Most cameras can fire the frames for you with auto exposure bracketing. Moving subjects such as leaves and people can leave ghosts; most merging software has a deghosting option. Many editors can also merge the frames into a single raw file that you then edit like any other.

Even when one raw frame would just about hold a scene, a bracket gives cleaner shadows, because in the bright frame they were recorded with plenty of light instead of lifted out of the noise. For work where that matters, bracket anyway.

The merged result is an HDR image. It then has to be squeezed back into the range a screen or print can show, which is tone mapping. Done gently it looks like what you saw. Pushed hard it gives the gray halos and flat, glowing look that gave HDR a bad name. That look comes from the tone mapping, not from the merge.

Would you rather blend the frames by hand, with full control of each area? Exposure blending covers combining exposures with masks instead of an automatic merge.

4. Add light to the shadows

Instead of holding back the highlights, raise the shadows. A reflector or fill flash lifts a backlit face toward the brightness of the sky behind it, shrinking the scene’s range at the source. A diffuser held between the sun and a close subject does the same job from the other end, taking the hardest light off the highlights. Both work for portraits and close subjects; they do nothing for a distant valley.

5. Wait for better light

The scene’s range changes through the day. Thin cloud, the low sun of the golden hour, or the blue hour, when the sun is just below the horizon before sunrise or after sunset, can bring a scene within reach of one exposure. At night, city lights are far brighter than the dark sky; in the blue hour, the sky catches up with them for a few minutes. Waiting is often the simplest answer.

Is the light the problem, not the camera? The golden hour calculator tells you when the low, soft light arrives where you are.

Dynamic range on film

Film has its own range, usually called latitude. Color negative film is forgiving of overexposure and can hold detail several stops above a normal exposure, which is why it handles bright skies well. Slide film is much less forgiving, closer to five stops of usable range, and needs careful exposure for the highlights. Digital sensors behave more like slide film at the top end: once highlights clip, they are gone. The zone system grew out of fitting a scene’s range onto film and then onto paper, and its way of thinking in stops still works with a digital camera. Film photography covers the differences between stocks.

Common mistakes with dynamic range

• Expecting the camera to see what you saw. Your eye adapts as it looks around; the camera makes one exposure. Check what the frame holds, not what you remember.
• Trusting the camera’s screen. A bright screen in a dark place makes everything look fine. Check the histogram and highlight warnings.
• Raising ISO in high contrast light. Base ISO holds the most range. Use a tripod and a slower shutter speed if the subject is still.
• Using a plain ND filter to fix a bright sky. It darkens everything equally. A graduated filter or bracketing is what you need.
• Clipping the highlights to keep the shadows clean. Shadows can be lifted; clipped highlights cannot be brought back.
• Lifting every shadow to full brightness. A scene with no dark areas looks flat. Keep some deep shadow for depth; dodging and burning lets you lift only the parts that need it.
• Buying range instead of using it. A stop more on a spec sheet matters less than base ISO, careful exposure and raw files.

Try this: make a stop map of a window

Ten minutes, any camera that can spot meter, or a phone camera app with a manual mode that shows the shutter speed it chooses. Stand inside a room and frame a window with a sunlit view and some of the dark wall around it.

1. Set base ISO, a fixed aperture such as f/8, and spot or tap metering. Meter the brightest thing outside that should keep detail, and note the shutter speed the camera suggests. Then meter the darkest part of the wall that should keep detail and note that speed.
2. Count the stops between the two speeds. From 1/1000 to 1/15 is 6 stops: 1/500, 1/250, 1/125, 1/60, 1/30, 1/15.
3. Now shoot one raw frame exposed so the view outside just stops blinking with highlight warnings on. Lift the wall in editing as far as it will go.

How to judge it: if the wall comes up clean and the view keeps its detail, the scene fits your camera, and you now know roughly how many stops apart two areas can be before it does not. If the wall turns to grainy mush, take two more frames at plus 2 and plus 4 stops and merge them, and compare. Do the same test on an overcast day and watch the gap shrink.

Photographing people in the street against bright and dark backgrounds? Using dynamic range in street portraits shows how to place a face against both.

Frequently asked questions about dynamic range

What is a good dynamic range for a camera?

Measured figures depend on the test, so compare cameras on the same one rather than on brochure numbers. In practice, a camera that holds around 12 stops at base ISO on an engineering test handles most daylight scenes, and bracketing covers the rest. Technique makes more difference than a stop between two bodies.

Does a higher ISO reduce dynamic range?

Yes. Raising ISO makes highlights in the scene clip sooner, so the widest range is always at base ISO. How much each stop of ISO costs depends on the sensor: close to a stop on some, less at low ISOs on others.

What does 14 stops of dynamic range mean?

It means the camera can record detail from its brightest unclipped tone down to a tone 14 stops darker. Each stop is a doubling, so that is a ratio of 2 multiplied by itself 14 times, 16,384 to 1. Whether those bottom stops look clean depends on how the figure was measured, as explained above.

What makes a high dynamic range sensor?

A high ceiling and a low floor: photosites that hold many electrons before they fill, and very low read noise, so faint detail is not lost in it. Dual gain designs widen the range at higher ISOs. Some sensors, common in phones, security and car cameras, go further and capture a short and a long exposure almost at the same moment, or read each pixel at two gains, and combine them on the chip; security cameras often sell this as wide dynamic range, or WDR. That is HDR done inside the sensor rather than in editing.

Is HDR the same as dynamic range?

No. Dynamic range is a measurement of a span of tones. HDR, high dynamic range, is a technique: merging several exposures so one image holds more range than one frame could. The result is then tone mapped so a screen or print can show it.

Why does my phone handle a bright sky better than my camera?

Your phone usually merges a burst of frames at different exposures without telling you, which is a form of HDR. A camera set to a single raw frame records one exposure. Bracket with the camera and merge the frames, and it can match or beat the phone.

Why does my photo lose detail in the sky?

The sky is much brighter than the land, often by more stops than one exposure can hold. If you exposed for the land, the sky clipped. Expose for the sky and lift the land in editing, use a graduated filter, or bracket and merge.

Test your knowledge

• Dynamic Range Quiz
• Sensor Physics Quiz
• HDR and Tone Mapping Quiz