White balance is the color correction that makes neutral objects appear neutral in photography by compensating for the color of the light that illuminates the scene. A camera sensor records light as it arrives, so a white wall lit by a tungsten bulb is recorded as orange and the same wall in open shade is recorded as blue. White balance scales the red, green and blue channels so that a gray or white reference ends up with equal channel values, removing the overall color cast from the image.
In practice, white balance is described with two coordinates: color temperature in Kelvin, which runs along the warm–cold axis, and tint, which runs along the green–magenta axis. The human visual system performs a similar adjustment automatically through chromatic adaptation, which is why scenes look normal to the eye while the uncorrected recording does not.
How white balance corrects the light source
White balance works by applying per-channel gain multipliers to the raw sensor data before the image is converted into an output color space. The camera estimates the color of the illuminant, then amplifies or reduces the red and blue channels relative to green until a neutral surface produces R = G = B. Green is usually the reference channel because a Bayer sensor has twice as many green photosites and green carries most of the luminance information.
The illuminant estimate can come from a preset, a manually entered Kelvin value, a custom measurement of a gray card, or the camera’s auto white balance algorithm. Auto algorithms analyze the scene for likely neutral areas, highlights and known light signatures; modern cameras also use face detection and scene recognition. All of these methods answer the same question: what color is the light, and how much must each channel change to cancel it.
Correction is applied once, globally, to the whole frame. This is the central limitation: a single white balance setting can neutralize only one illuminant. When a scene mixes daylight from a window with warm interior lamps, any setting leaves at least one area tinted.
Color temperature and tint values
Correlated color temperature describes a light source by comparing it with an ideal black-body radiator. A low value such as 2700 K means reddish, warm light; a high value such as 7500 K means bluish, cold light. The camera setting states the temperature of the light being corrected, so setting 3200 K under tungsten light makes the camera add blue, and setting a higher value than the real light makes the image warmer.
The following table lists typical light sources with approximate correlated color temperatures and the matching camera preset. Real values vary with lamp type, age, time of day and atmosphere, so the figures are starting points rather than fixed constants.
| Light source | Approximate color temperature | Matching preset | Uncorrected appearance |
|---|---|---|---|
| Candle flame | 1800–2000 K | Custom or manual Kelvin | Deep orange |
| Household tungsten bulb | 2700–3000 K | Tungsten or incandescent | Orange-yellow |
| Studio halogen lamp | 3200–3400 K | Tungsten | Mild yellow |
| Midday direct sunlight | 5200–5600 K | Daylight | Neutral |
| Electronic flash | 5500–6000 K | Flash | Neutral to slightly blue |
| Overcast sky | 6000–7000 K | Cloudy | Bluish gray |
| Open shade under blue sky | 7000–10000 K | Shade | Strong blue |
Many artificial sources do not sit on the black-body curve. Fluorescent tubes and low-quality LEDs often add a green component that no Kelvin value can remove, which is why the separate tint control exists. A positive tint adds magenta to cancel green, and a negative tint adds green to cancel magenta.
White balance methods compared
Each method trades convenience against accuracy and consistency. The table summarizes how the main approaches behave in practice.
| Method | How it works | Strength | Weakness |
|---|---|---|---|
| Auto white balance | Algorithm estimates the illuminant per frame | Fast and usually accurate in daylight | Shifts between frames; struggles with dominant colors |
| Presets | Fixed values for common sources | Consistent across a series | Approximate; ignores real lamp variation |
| Manual Kelvin | Photographer enters the temperature | Precise and repeatable | Requires knowing or metering the light |
| Custom gray card reading | Camera measures a neutral target | Most accurate for one illuminant | Must be repeated when the light changes |
| Correction in post-processing | Eyedropper or sliders on a RAW file | Fully reversible with RAW | Limited and lossy on JPEG files |
Auto white balance tends to fail in predictable situations. A frame filled with one dominant color, such as green foliage or a red stage backdrop, can be misread as a color cast and “corrected” toward its complement. Warm scenes such as sunsets and candlelit interiors are often over-corrected, because the algorithm cannot know that the warmth is the subject.
White balance in RAW and JPEG workflows
The file format decides whether white balance is a creative decision or a permanent one. In a RAW file, the camera only stores the white balance multipliers as metadata; the sensor data stays untouched, so the setting can be changed later without any quality loss. In a JPEG file, the correction is baked into the 8-bit image together with the tone curve and the camera color profile.
Correcting a JPEG afterward means stretching already compressed channels. Moderate shifts work, but large corrections cause posterization in skies, clipped channels and noise in the channel that was amplified most. For this reason, photographers who shoot JPEG or video should set white balance accurately at capture, while RAW shooters can treat it as an editing parameter.
This decision table shows how to approach white balance under common shooting conditions.
| Situation | Recommended approach | Reason |
|---|---|---|
| Event series under stable light | Custom gray card reading or fixed Kelvin | Identical color across hundreds of frames |
| Changing outdoor light, RAW capture | Auto white balance, refine in editing | Speed at capture, precision later |
| Video or multi-camera production | Manual Kelvin matched on every camera | Color cannot be fixed per frame afterward |
| Sunset or candlelight | Daylight preset or a fixed warm value | Preserves the intended warm mood |
| Product or reproduction work | Gray card plus color checker profile | Accurate, measurable neutrals |
Mixed light and common mistakes
Mixed lighting is the most difficult white balance problem because the frame contains several illuminants at once. Professionals solve it at the source by gelling lights: a CTO gel (color temperature orange) warms a flash to match tungsten, and a CTB gel (color temperature blue) cools a tungsten lamp toward daylight. Once all sources match, a single white balance setting works for the whole frame. In editing, local adjustments can neutralize individual areas, but this is slow and rarely perfect.
- Confusing the slider direction: a higher Kelvin setting makes the image warmer, because it tells the camera the light is bluer than it is.
- Judging color on an uncalibrated screen: a monitor with its own cast misleads every correction; use a calibrated display.
- Neutralizing everything: golden-hour warmth and blue-hour blueness are often the reason for the photograph.
- Ignoring tint: fluorescent and LED sources often need a magenta shift, not a Kelvin change.
- Relying on auto white balance for a sequence: frame-to-frame shifts make a series look inconsistent.
Stripped to the essentials
Every light has its own color. Light bulbs are orange, shade is blue and fluorescent tubes can be slightly green. Your eyes adjust to this without you noticing, so a white shirt always looks white. A camera does not adjust by itself, so it needs to be told what color the light is.
White balance is that instruction. It tells the camera which color counts as plain white, and the camera shifts all colors so that white looks white again. Get it right and skin, food and walls look natural; get it wrong and the whole picture looks tinted. You can also leave some warmth on purpose when the mood calls for it.
Where it is applied
A wedding photographer covers a reception in a hall lit by warm tungsten chandeliers at about 2800 K, while the bounced flash on the camera produces light at about 5500 K. Without correction, the guests lit by flash look cold and the background glows orange, and no single setting can fix both.
The photographer attaches a full CTO gel to the flash, which brings its output close to tungsten, then sets the camera to 3000 K and takes a custom reading from a gray card at the start of the evening. The photographs are shot in RAW, so the setting can be refined later.
The result is consistent, natural skin tones across several hundred frames, with the background candles still slightly warm. Editing time drops because one correction can be synchronized across the entire series instead of fixing each frame individually.