Short answer
Each colour channel has a ceiling. When the brightest channel of a colour reaches it and stops rising while the others keep climbing, the ratio between channels — which is what defines hue — changes. A bright blue sky loses its blue lead and slides towards cyan or white; overexposed skin loses red and turns yellow; saturated red flowers flatten into orange blobs. The fix is exposure that keeps every channel below its limit, not editing after the fact.
A colour's hue is determined by the proportions of red, green and blue. A saturated sky might record, in linear sensor terms, a large blue value, a moderate green and a small red. Increase exposure and all three rise together, keeping their ratio, until blue hits the sensor's full-well or the file's maximum. From then on blue stays fixed while green and red continue to rise. The ratio shifts towards green — cyan — and as green also clips the patch heads for white. Nothing has gone wrong with the sky; the recording has run out of room in one channel first.
A neutral grey spreads its signal evenly, so all three channels hit the ceiling at the same exposure. A saturated colour concentrates its signal in one or two channels, which reach the limit while the overall image still looks well exposed. That is why a bright red rose, a yellow jacket or a blue neon sign can lose detail and hue while the grey pavement beside it is fine. The luminance histogram most cameras show by default averages the channels and hides this; the per-channel RGB histogram shows it clearly.
Skin clips towards yellow because skin's red channel is the strongest; once red saturates, rising green pulls the ratio towards yellow.
Even before a channel clips at the sensor, a JPEG's tone curve compresses the top of the range to fit a scene into 8 bits. Applied to each channel separately, a steep shoulder squeezes the strongest channel more than the weaker ones, desaturating and shifting bright colours towards the hue of whatever channel is still climbing. Modern raw converters and phone pipelines try to preserve hue in highlights by compressing brightness rather than each channel independently, but the underlying clipping at capture cannot be undone.
In a raw file, if only one channel clipped, a converter can estimate the missing channel from the other two and from neighbouring pixels — highlight reconstruction. It is a guess, usually a plausible one for near-neutral highlights and a poor one for saturated colours. If all three channels clipped, there is no information left: the area is white in every version of the file. Exposing for the highlights, using a lower exposure in bright sun, or merging a bracketed or burst sequence are the only real remedies.
Why: Blue channel clipped; green keeps rising.
Fix: Reduce exposure until the blue channel histogram stops touching the right edge.
Why: Red channel clipped on the brightest skin.
Fix: Expose for the face, add fill light or move into open shade.
Why: Red channel saturated long before the overall image looks bright.
Fix: Underexpose by a stop or more and check the red histogram.
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
FactStrong evidence
A published burst pipeline sets each frame's exposure low enough to avoid blowing out highlights and then brightens shadows by merging frames, because highlights lost at capture cannot be recovered.
FactModerate evidence
Hue in an RGB image is set by the ratio of the channels, so saturation of one channel before the others shifts the recorded hue towards the channels that are still rising.
Caveat: The CIE and IEC documents define the colour spaces; the clipping behaviour follows from those definitions rather than being stated in them.
Source: CIE 015:2018 Colorimetry, 4th edition; International Electrotechnical Commission (IEC) webstore and catalogue
Colourwise interpretationModerate evidence
Luminance histograms can show a well-exposed image while a single saturated colour channel is clipped; per-channel histograms are the more reliable exposure check for colourful subjects.
Based on: Follows from the channel-ratio mechanism above: an average of three channels stays below the maximum while one channel sits at it.
Reviewed 29 September 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.