Dynamic range in photography is the ratio between the brightest signal a sensor can record before clipping and the darkest signal it can still distinguish from noise. It is usually expressed in stops (EV), where each stop is a doubling of light, so a sensor with 12 stops of dynamic range separates tones across a brightness ratio of about 4096:1 in a single exposure.
As an image quality measure, dynamic range decides how much detail survives in bright skies and deep shadows at the same time. It depends on two limits: full-well capacity at the top and the noise floor, dominated by read noise, at the bottom. The term is also used for scenes, where it describes the brightness range that the camera must capture.
How sensor dynamic range is determined
Sensor dynamic range is defined as the ratio of maximum signal to noise floor. The upper limit is the full-well capacity, the number of electrons a photosite can hold before it saturates; beyond it, highlights clip to pure white. The lower limit is the level at which the signal equals the noise, set mostly by read noise from the sensor electronics, plus dark current during long exposures.
In stops, the calculation is DR = log2(full-well capacity ÷ read noise). A photosite holding 50,000 electrons with 3 electrons of read noise gives log2(16,667), or about 14 stops. In decibels, the same ratio is 20·log10, roughly 6 dB per stop. Measured “engineering” dynamic range uses a signal-to-noise ratio of 1 as the lower bound, while “photographic” dynamic range uses a stricter threshold such as SNR 20, so published figures differ by one to three stops depending on the definition.
Two further factors shape the result. The bit depth of the analog-to-digital converter caps how finely tones are encoded, so a 12-bit file cannot store more than about 12 stops of linear data. Pixel size matters less than it appears: larger photosites hold more electrons, but when output size is normalized, sensor area and read noise performance dominate.
Typical dynamic range values
The table compares the approximate dynamic range of common capture and display media at their best settings. Values are indicative ranges, because measurement methods, ISO setting and processing all move the figures.
| Medium | Approximate dynamic range | Main limiting factor |
|---|---|---|
| Smartphone sensor, single frame | 9–11 stops | Small photosites and limited full-well capacity |
| APS-C camera at base ISO | 12–14 stops | Read noise and ADC bit depth |
| Full-frame camera at base ISO | 13–15 stops | Read noise at the shadow end |
| Color negative film | 12–14 stops | Gradual shoulder and toe of the film curve |
| Color slide film | 5–7 stops | Steep contrast of reversal emulsion |
| SDR display or 8-bit JPEG | About 6–8 stops | Gamma encoding and display contrast |
| HDR display | 10 or more stops | Peak brightness and black level |
A real scene can exceed all of these. A sunlit landscape with a dark forest interior or a room with a bright window can span 15 to 20 stops, which is why photographers need techniques beyond a single exposure.
Dynamic range, ISO and noise
Dynamic range is highest at base ISO and falls as ISO rises. Each doubling of ISO amplifies the signal so that highlights clip one stop sooner, while the noise floor drops by less than a stop; the net effect is a loss of roughly one stop per ISO doubling once read noise is no longer the limiting factor. At high ISO, shot noise from the light itself dominates and the range narrows further.
Many current sensors use dual conversion gain, switching to a high-gain readout mode at a set ISO, often between ISO 400 and ISO 800. This lowers read noise and briefly recovers dynamic range. Sensors with very low read noise are called ISO-invariant: underexposing at base ISO and brightening later produces nearly the same result as raising ISO in camera, which protects highlights.
The table shows how dynamic range relates to neighboring image quality concepts in the same field.
| Concept | What it describes | Relation to dynamic range |
|---|---|---|
| Signal-to-noise ratio | Signal strength relative to noise at a given tone | Dynamic range is the span where SNR stays usable |
| Read noise | Noise added by sensor readout electronics | Sets the shadow limit at low ISO |
| Shot noise | Random variation in photon arrival | Limits shadow quality as light decreases |
| Luminance noise | Brightness grain in the image | Visible symptom of a reduced noise floor margin |
| Contrast | Tonal difference within a rendered image | A rendering choice, not a capture capacity |
Capturing scenes beyond the sensor range
When the scene dynamic range exceeds the sensor, the photographer must decide which tones to sacrifice or extend the capture. The standard approach with a single frame is expose to the right: set exposure so that the brightest important highlight sits just below clipping, then lift shadows in RAW processing, where the extra shadow signal reduces noise.
This decision table summarizes the common techniques and when each applies.
| Technique | Best use | Main drawback |
|---|---|---|
| Single RAW exposure, highlights protected | Moderate contrast up to about 13 stops | Noise in strongly lifted shadows |
| Exposure bracketing with HDR merge | Static landscapes and interiors | Ghosting with moving subjects |
| Graduated neutral density filter | Bright sky over a darker foreground | Straight transition crosses objects on the horizon |
| Fill flash or reflector | Backlit portraits | Limited range and visible light direction |
| Multi-frame computational merge | Smartphones and handheld scenes | Processing artifacts and less control |
Capturing the range is only half the task. The output medium usually shows fewer stops than the file contains, so tone mapping compresses the captured range into the display range. Heavy compression produces a flat, halo-prone look, which is a rendering problem rather than a sensor problem.
Common misconceptions about dynamic range
- More bits do not create more dynamic range: a 14-bit file only encodes the sensor output more finely; the sensor still sets the limit.
- Manufacturer figures are not comparable: video specifications often quote stops at a lenient SNR threshold, while still-image tests use stricter criteria.
- Clipped highlights cannot be recovered: once all channels saturate, there is no data left; highlight recovery works only when at least one channel retains detail.
- Shadow lifting is not free: the deepest recoverable shadows always carry more noise and less color accuracy than midtones.
- Dynamic range is not contrast: a low-contrast rendering can come from a high-dynamic-range file, and the reverse is also true.
The idea in brief
Imagine photographing a room with a sunny window. Your eyes see both the bright view outside and the details of the dark sofa. A camera often cannot: either the window turns pure white or the sofa turns solid black.
Dynamic range is the width of that window of brightness the camera can handle at once. A camera with more dynamic range keeps detail in both the bright and dark parts of the same photo. The wider the range, the more forgiving the camera is in harsh, high-contrast light.
A practical scenario
A real estate photographer shoots a living room at midday. The interior walls meter at 1/15 s at f/8, while the garden visible through the window meters at 1/1000 s, a difference of six stops on top of the normal tonal spread, so the scene spans roughly 16 stops in total.
Working at base ISO 100 on a tripod, the photographer brackets five RAW frames two stops apart, from 1/4000 s to 1/15 s, and merges them into a single HDR file. Tone mapping is kept moderate so the walls stay natural and the garden remains green instead of white.
The final image shows detail in both the room and the view, without the clipped window that a single frame would produce. A single exposure would have forced the choice between a dark room and a blank white window.