Highlight recovery is the process of pulling detail back from overexposed regions of a raw file during editing, typically using the highlight slider in Lightroom, Capture One, ACR, or any equivalent raw processor. Modern raw converters can recover roughly one to two stops of detail above the apparent clipping point, depending on the camera, the ISO, and how cleanly the individual color channels clipped. Beyond that range, the data is genuinely lost and no slider will bring it back.
Recovery works because raw files store more bit depth than JPEG and because the three color channels rarely clip simultaneously. When the red channel saturates first (as it often does on skies near the sun or on bright skin), the green and blue channels still hold structure. The processor can rebuild a plausible luminance from the unclipped channels, dropping the recovered highlights into a tonal range below pure white. The technique is sometimes called luminance inference or single-channel recovery.
JPEG files behave differently. Their tone curve has already been applied and their dynamic range compressed to 8 bits per channel before the file was written, so any clipped highlight in a JPEG was already discarded at capture. Pulling down highlights in JPEG only gray-tones the already-white pixels, producing flat patches that do not gain real detail. This is one of the strongest practical arguments for shooting raw, especially in high-contrast scenes.
The technique that pairs with highlight recovery is exposing to the right, abbreviated ETTR: deliberately pushing the exposure as far toward the right edge of the histogram as the scene allows without true clipping, then recovering in post. Brighter exposures capture more signal relative to read noise, and the recovered highlights then look cleaner than a properly exposed midtone pushed up in editing. Many landscape and architectural photographers default to this method, watching the blinkies and the raw histogram rather than the JPEG preview.
Recovery has limits. Specular highlights (the sun, light bulbs, mirror glints) are usually clipped beyond rescue and look fine that way; trying to recover them tends to introduce odd magenta or cyan tints from channel imbalance. Skin highlights near saturation can pull back cleanly but turn rubbery if pushed too hard, since the texture has been crushed at the sensor. Skies often recover impressively, restoring cloud structure that looked entirely white in the embedded preview.
Cameras with higher native dynamic range and ISO invariance recover more cleanly because the underlying signal is less compressed and the noise floor is lower. Sony, Fujifilm, Nikon, and the newest Canon sensors all give very usable recovery; older CMOS and CCD sensors are stingier. The safest workflow is to learn your camera’s actual recovery margin by deliberately overexposing test frames in stops above the meter, then seeing where the slider stops finding real data.