Highlight recovery is a group of image processing techniques in photography that restore visible detail in the brightest areas of an image, such as skies, clouds, white clothing or specular surfaces, by drawing on data that the default rendering would have pushed to pure white. It works best on RAW files, which typically hold extra headroom above the point where a standard JPEG rendering clips.
Highlight recovery can bring back only what the sensor actually recorded, plus a limited amount of reconstructed information. Once all three color channels are saturated at a pixel, the original detail is gone, and any result there is an estimate.
Why recoverable highlight data exists
Recoverable highlights exist because a raw file stores linear sensor data with more range than a typical rendered image displays. The default tone curve and white balance in a raw converter map scene values to output in a way that often clips the top part of the data, even though the sensor did not saturate there. Pulling those values back down reveals detail that was recorded but not shown.
The amount of headroom depends on the camera, the raw converter’s baseline exposure and the scene. It typically ranges from a fraction of a stop to about one or two stops above the brightness at which the default rendering appears white. JPEG files, by contrast, are already rendered and quantized to 8 bits per channel, so they hold little or no data above their white point.
| File type | Bit depth | Typical highlight headroom | Recovery potential |
|---|---|---|---|
| RAW, 14-bit | 14 bits per channel | Often about 1 to 2 stops | High |
| RAW, 12-bit | 12 bits per channel | Similar headroom, coarser steps | High |
| HEIC or 10-bit HEIF | 10 bits per channel | Small | Limited |
| JPEG | 8 bits per channel | Little or none | Very limited |
How highlight recovery algorithms work
Highlight recovery combines three distinct operations. The first is tone compression, which lowers the brightest tones with a highlights slider or tone curve so that unclipped detail becomes visible. The second is channel reconstruction, which rebuilds a clipped color channel from channels that still hold data. The third is inpainting or synthesis for regions where all channels are saturated.
Channel reconstruction is possible because the three channels saturate at different scene brightness levels. Under daylight, the green channel usually clips first, since it is the most sensitive after white balance. Software estimates the missing green values from the ratios observed in nearby unclipped pixels and from the remaining red and blue data, which often recovers believable texture in clouds and skin.
| Method | What it uses | Strength | Typical artifact |
|---|---|---|---|
| Tone compression | Unclipped data above the default white | Natural, accurate detail | Flat, gray highlights if overdone |
| Channel reconstruction | Unclipped channels and neighbors | Recovers partly clipped areas | Color shifts, often magenta or cyan |
| Clip to white or desaturation | Clipped channels set neutral | Clean, avoids color casts | Loss of color in bright areas |
| Inpainting or AI synthesis | Surrounding image content | Fills fully clipped regions | Plausible but invented detail |
Highlight recovery and related controls
Highlight recovery overlaps with several other processing controls, and confusing them leads to poor results. The highlights slider in most raw converters applies tone compression locally, while the whites slider sets the white point, and exposure shifts the entire image.
| Control | What it changes | When to use it |
|---|---|---|
| Exposure | Overall brightness linearly | Whole image too bright |
| Highlights slider | Upper tonal range, often locally | Bright areas lacking detail |
| Whites slider | Position of the white point | Setting the brightest acceptable tone |
| Tone curve | Precise tonal mapping | Fine shaping of the roll-off |
| Local adjustments | Selected regions only | Bright sky over a correctly exposed scene |
Recovery is closely related to highlight clipping, which describes the problem, and to HDR photography, which avoids it by merging several exposures. A single raw file with recovery handles moderate contrast; extreme contrast still needs bracketing or added light.
Recognizing what can and cannot be recovered
Before editing, a photographer should check whether highlights are clipped in the raw data or only in the preview. Camera histograms and blinking warnings are based on the embedded JPEG, so they usually warn earlier than the raw file actually clips. Raw converters with per-channel clipping warnings show the true situation.
- One channel clipped: usually recoverable with reconstruction and modest color correction.
- Two channels clipped: partial texture may return, with a high risk of color errors.
- All three channels clipped: no original detail; only neutral fill or synthesis remains.
- Specular highlights, such as sun reflections on metal or water, are normally left white because viewers expect them to be.
Strong recovery also lowers local contrast in bright areas. Skies may turn gray and lifeless, and bright skin can look muddy. Restoring contrast with clarity or a tone curve after recovery usually gives a more natural result.
Capture habits that make recovery easier
The most reliable highlight recovery begins at capture. Exposing to the right maximizes shadow quality but must stop short of raw clipping, so experienced photographers learn how much margin their camera’s highlight warning leaves and sometimes use a flatter picture style to make the preview more representative.
| Scene | Capture approach | Reason |
|---|---|---|
| Bright sky with dark foreground | Protect sky, lift shadows later | Raw shadows tolerate brightening |
| White wedding dress in sun | Meter for the dress | Fabric texture clips quickly |
| Stage lights in frame | Accept clipped light sources | Sources are expected to be white |
| Extreme contrast landscape | Bracket exposures | Exceeds single-frame range |
A non-technical view
In a photo with a bright sky, the clouds sometimes look like plain white patches with no shape. Highlight recovery is a way of pulling those bright areas back down so the shapes and textures reappear.
It works because a RAW photo quietly stores more information than the screen first shows. But there is a limit. If the camera recorded only pure white, there is nothing left to bring back, and software can only guess.
A concrete example
A wedding photographer shoots the couple leaving a church into midday sun. The exposure is set for the groom’s dark suit, and the bride’s white dress and the sky look completely white on the camera screen.
In the raw converter, per-channel warnings show that the dress is clipped only in the green channel, while the sky is partly clipped in two channels. The photographer reduces the highlights slider by about 70 and lowers whites slightly, then adds a gentle local adjustment to the sky. The lace pattern of the dress returns clearly, and the sky regains soft blue tones, while a small area near the sun stays white and is left untouched.