Auto white balance (AWB) is the camera function that estimates the color of the light illuminating a scene and applies a correction so that neutral objects, such as white paper or gray concrete, are rendered without a color cast. In photography it is the default white balance setting on almost every camera and smartphone, and a central concept of the Color & white balance topic.
AWB is an estimation problem: the camera sees only the mixture of light color and surface color and must guess which part belongs to the light. Modern systems solve it well in ordinary scenes, but they can misjudge scenes dominated by one color, mixed lighting and deliberate atmospheric light such as sunsets or candlelight.
How auto white balance estimates the light
Auto white balance analyzes the image signal, estimates the color of the illuminant and then scales the red, green and blue channels so that the estimated light becomes neutral. The correction is expressed as channel gains, usually with green as the reference and red and blue amplified or reduced.
Classic algorithms use simple assumptions. The gray world assumption expects the average of a scene to be neutral gray, and the white patch assumption expects the brightest area to be white. Both fail in predictable ways: a green lawn pushes a gray world estimate toward magenta, and a bright yellow sign misleads a white patch estimate.
Current cameras combine statistics from many image zones with data from the exposure and metering system, face and scene detection, and in some models a separate color or flicker sensor. Many recent systems also use machine learning models trained on large sets of labeled images to recognize typical light sources and scene types.
The table below summarizes the main estimation approaches and where each tends to fail.
| Approach | Core assumption | Typical failure |
|---|---|---|
| Gray world | Average scene color is neutral | Scenes dominated by one color |
| White patch | Brightest area is white | Bright colored objects or clipped highlights |
| Gamut and statistics based | Possible colors depend on the illuminant | Unusual or artificial light spectra |
| Learning based | Scene patterns predict the light | Scenes unlike the training data |
Color temperature range and tint correction
AWB corrects along two axes: color temperature in kelvin, the blue to amber axis, and tint, the green to magenta axis. Most cameras operate reliably within roughly 2500 to 7500 K, and many restrict the automatic range at the extremes.
| Light source | Approximate color temperature | Typical AWB result |
|---|---|---|
| Candlelight | About 1800 to 2000 K | Remains warm, often intentionally |
| Tungsten bulb | About 2700 to 3200 K | Mostly corrected, slight warmth may remain |
| Daylight at midday | About 5000 to 5600 K | Accurate in most cameras |
| Overcast sky | About 6000 to 7000 K | Corrected, sometimes slightly cool |
| Open shade | About 7000 to 9000 K | Can remain bluish |
| Fluorescent and LED | Varies widely, often with green shift | Tint errors more common than kelvin errors |
Artificial sources with discontinuous spectra, such as some LEDs and sodium lamps, are especially difficult, because no white balance correction can fully restore colors that the light does not contain. AWB then produces a neutral average, but individual colors may still look wrong.
AWB variants and priority settings
Many cameras offer several AWB variants that decide how much of the original light color to keep. This reflects a real choice: perfect neutrality is not always the most natural-looking result.
| Setting | Behavior | When it helps |
|---|---|---|
| Ambience priority | Keeps some warmth under tungsten light | Interiors and evening scenes with mood |
| White priority | Neutralizes warm light more strongly | Product, document and interior work |
| Natural light setting | Tuned for daylight and outdoor mixes | Landscapes and travel |
| AWB with shift | Applies a user offset to every estimate | Matching a preferred house look |
A white balance shift applies a fixed bias along blue to amber and green to magenta, and it combines with AWB rather than replacing it. Some cameras also allow an AWB lock for video, which prevents visible color changes when the framing moves.
Auto white balance compared with other white balance methods
Auto white balance is one of several methods for setting white balance, and each trades convenience against consistency. AWB adapts to every frame, which is useful when light changes but problematic when a series must match.
| Method | Consistency | Accuracy | Effort |
|---|---|---|---|
| Auto white balance | Varies frame to frame | Good in typical scenes | None |
| Presets | Constant | Approximate | Low |
| Kelvin setting | Constant | Good if value is known | Moderate |
| Custom white balance | Constant | High for the measured light | Requires a gray card or target |
| Post-processing on RAW | Fully adjustable | High with a reference target | Editing time |
With RAW files, the AWB setting is stored only as metadata and can be changed later without quality loss. With JPEG or HEIC files, the correction is baked into the image, so a large AWB error is harder to repair.
Where auto white balance goes wrong
AWB errors follow recognizable patterns, and knowing them helps photographers decide when to switch to a fixed setting.
- Dominant colors, such as a red wall, green forest or blue swimming pool, are interpreted as colored light and partly removed, leaving a complementary cast.
- Sunsets and golden hour are often neutralized, which removes the warm mood the photographer wanted; a daylight preset keeps it.
- Mixed lighting, for example window light combined with tungsten lamps, leaves one area blue and the other amber, because a single global correction cannot fix both.
- Inconsistent series appear when framing changes slightly between shots, which complicates batch editing of events and products.
- Flicker from artificial lights at fast shutter speeds can shift color between frames in a burst.
Put simply
Light is not always the same color. Indoor bulbs are orange, shade is bluish and noon sun is close to neutral. Our eyes adjust automatically, so a sheet of paper looks white everywhere, but a camera needs help.
Auto white balance is the camera trying to do what your eyes do. It guesses the color of the light and corrects the photo so white things look white. It usually guesses well, but a scene full of one strong color can fool it.
In practice
A wedding photographer covers a reception in a hall lit by warm tungsten chandeliers, with daylight from tall windows in the late afternoon. Using auto white balance, she notices that consecutive images of the same table shift between cooler and warmer renderings as the framing changes from window side to room side.
For the speeches she switches to a fixed 3200 K setting and photographs a gray card under the chandeliers, shooting RAW. During editing she applies one correction to the whole sequence, and the series matches from frame to frame, while the earlier AWB images need individual adjustment.