Color depth

On this page 7

Color depth, also called bit depth, is the number of bits used to store the value of each color channel in a pixel, and it determines how many distinct tonal and color steps an image can represent. In photography it is a core measure of image quality, because it decides whether smooth gradients such as skies and skin stay continuous or break into visible bands after editing.

Color depth is usually stated per channel (8-bit, 12-bit, 14-bit, 16-bit) or per pixel as the sum of three channels (24-bit or 48-bit). The two conventions describe the same thing, and confusing them is one of the most common sources of misunderstanding in camera specifications and file format comparisons.

How color depth quantizes tones

Color depth works by dividing the tonal range into discrete levels. A channel with n bits can hold 2 to the power of n values, so each additional bit doubles the number of available steps between black and full saturation. This process is called quantization, and every digital image, whether captured by a sensor or created in software, passes through it.

In a camera, the analog-to-digital converter (ADC) measures the voltage from each photosite and assigns it a whole number. A 14-bit ADC produces values from 0 to 16,383, while an 8-bit output offers only 0 to 255. The finer the scale, the smaller the difference between adjacent levels, and the more editing the data can tolerate before steps become visible.

The table below shows how the number of levels grows with bit depth and where each depth typically appears in a photographic workflow.

Bit depth per channelLevels per channelTotal colors (three channels)Typical use
8-bit256About 16.7 millionJPEG, web display, most monitors
10-bit1,024About 1.07 billionHEIC, HDR video, wide-gamut displays
12-bit4,096About 68.7 billionRaw files from many cameras and phones
14-bit16,384About 4.4 trillionRaw files from most interchangeable-lens cameras
16-bit65,536About 281 trillionTIFF and PSD editing working space
Levels per channel at common bit depths

Color depth, dynamic range and noise

Color depth is not the same as dynamic range, although the two are often confused. Dynamic range is the ratio between the brightest and darkest signal a sensor can record above its noise floor, while color depth is how finely that range is subdivided. A 14-bit file does not automatically capture 14 stops; it only provides enough levels to encode up to about 14 stops without losing information.

In practice, sensor noise limits useful bit depth. When read noise and shot noise exceed the size of one quantization step, the lowest bits contain mostly random variation rather than real detail. This is why a sensor with about 12 stops of engineering dynamic range gains little from a 16-bit ADC: the extra bits record noise more precisely, not more scene information.

Noise also plays a useful role. A small amount of random variation acts as natural dithering, which breaks up the hard edges between levels and hides banding. For this reason a slightly noisy 8-bit image can look smoother than a perfectly clean one rendered at the same depth.

The following comparison separates color depth from the neighboring image quality concepts it is most often mistaken for.

ConceptWhat it measuresUnitMain visible effect
Color depthNumber of levels per channelBitsSmoothness of gradients
Dynamic rangeSpan from darkest to brightest usable signalStops (EV)Detail in shadows and highlights
Signal-to-noise ratioSignal strength relative to noiseDecibels or ratioGrain and clarity of tones
Color gamutRange of reproducible hues and saturationColor space (sRGB, Adobe RGB)How vivid colors can appear
Color depth compared with related image quality terms

Bit depth in raw and processed files

A raw file stores linear sensor data, usually at 12 or 14 bits, before demosaicing and tone mapping. Because raw values are linear, half of all levels describe the brightest stop, and each darker stop receives half as many levels as the one above it. The deepest shadows therefore have relatively few steps, which is why aggressive shadow lifting reveals posterization and noise first.

Processed formats apply a gamma or tone curve before reducing bit depth. This redistributes levels toward the darker tones, where human vision is more sensitive to small differences, so an 8-bit gamma-encoded JPEG can look smooth on screen even though it holds far fewer levels than the original raw file. The cost appears later: an 8-bit file has little headroom for further editing.

Some cameras offer lossless, lossy and reduced-bit raw modes. A 12-bit mode can increase burst speed or reduce file size, and the visible difference from 14-bit is usually small at moderate ISO, but it can matter in heavily lifted shadows at base ISO.

Choosing the right color depth

The right color depth depends on how much the image will be edited and where it will be shown. The general rule is to capture and edit at high depth, then reduce depth only for final delivery. Converting an 8-bit file to 16-bit does not restore lost levels; it only prevents further losses from subsequent edits.

The decision table below matches common situations to a suitable depth and explains the reasoning.

SituationRecommended depthReason
Landscape with smooth sky14-bit raw, 16-bit editingGradients survive strong curves
High-volume event shooting12-bit raw or high-quality JPEGFaster bursts and smaller files
Web or social media delivery8-bit sRGB exportUniversal support, small size
Print in wide-gamut color space16-bit TIFFPrevents banding in saturated tones
HDR display output10-bit or higherWider brightness range needs more levels
Recommended color depth by situation

Working in a wide color space such as ProPhoto RGB at 8 bits is a frequent error. The larger gamut spreads the same 256 levels across a wider range of colors, which increases the step size and makes banding more likely.

Banding, posterization and other failure modes

When color depth is insufficient, the result is banding or posterization: visible stripes or flat patches where a smooth transition should be. These artifacts appear most often in clear skies, studio backdrops, skin tones and vignetted corners, where brightness changes slowly across many pixels.

The table below lists typical symptoms, their usual causes and practical fixes.

SymptomLikely causeFix
Stripes in a blue sky8-bit file edited with strong curvesRe-edit from the raw file in 16-bit
Blocky shadow tonesShadows lifted several stopsExpose to the right, use 14-bit raw
Banding only on one monitor8-bit or poorly calibrated displayCheck on a 10-bit calibrated screen
Bands after exportConversion to 8-bit without ditheringAdd slight noise before export
Common color depth problems and remedies

Not every band comes from the file itself. Display pipelines, compression in web platforms and graphics drivers can all introduce banding, so it is worth confirming the problem on a second device before reprocessing an image.

Explained without jargon

Think of color depth as the number of steps on a staircase that runs from black to the brightest color. With few steps, each one is tall and you can clearly see where one ends and the next begins. With many steps, the staircase feels almost like a smooth ramp.

An ordinary JPEG photo has 256 steps for each color, which looks fine as it is. A camera raw file has thousands of steps. That extra detail does not show at first, but it matters when you brighten or recolor a photo. More steps mean smoother skies and skin after heavy editing, while too few steps leave visible stripes.

A typical use case

A landscape photographer shoots a coastal sunrise with a clear, gradually brightening sky. To protect the bright horizon, the exposure leaves the foreground about two stops too dark. The camera records 14-bit lossless raw files, giving 16,384 levels per channel.

In editing, the photographer lifts the shadows by two stops and adds a strong S-curve to deepen the sky color. The raw file is processed into a 16-bit TIFF in Adobe RGB for retouching. A test copy made from an 8-bit JPEG of the same frame shows clear stripes across the sky after identical adjustments.

The final image is exported twice: a 16-bit TIFF for a large gallery print and an 8-bit sRGB JPEG with slight added grain for the web. Both versions show a smooth, band-free gradient, because the heavy editing was done while the full color depth was still available.