Clipping

Clipping is what happens when part of a photo is brighter or darker than the camera, the file or the screen can record, so that whole area is stored as the same flat value: pure white, pure black, or one color channel stuck at its limit. Texture, gradation and color in that area are gone, and in most cases nothing in editing can bring them back.

It matters because clipping is the one exposure mistake that removes information rather than just moving it. A photo that is a little too dark or too bright can usually be fixed; a sky that has clipped to white cannot be given its clouds back. Knowing where clipping happens, which kind matters and which does not, is what lets you expose with confidence in high contrast light.

What clipping is: a ceiling and a floor

Every pixel in a digital photo is a set of numbers, one each for red, green and blue. Each number has a maximum. In an 8-bit JPEG the maximum is 255; in a 12-bit raw file it is 4,095; in a 14-bit raw file it is 16,383. When more light arrives than the maximum can describe, the pixel does not record “more”; it records the maximum and stops. A cloud that was twice as bright as the ceiling and a cloud that was ten times as bright end up as the same number, and so does every bit of texture between them.

That is why a clipped area looks flat and pasted on. It is not just very bright; it is a patch where every pixel has the same value, so there is no edge, fold or tone left to see. The same thing happens at the other end: anything darker than the floor is stored as zero, or as noise with no picture in it.

The ceiling is hard. Each photosite on the sensor holds only so much charge, and once it is full, extra light changes nothing. Film behaves differently: its density rises more and more slowly as the light increases, so very bright areas fade gradually toward white over a stop or more. Digital clipping has no such soft shoulder in the raw data, which is why a blown sky often ends in a visible border instead of a gentle fade. The camera’s JPEG rendering and raw editors add a gentler curve on top, but the hard limit underneath is still there.

Is a bright photo always a clipped one? No, and the difference matters: overexposure sorts out bright, overexposed and clipped.

The four places a photo can clip

Clipping is not one event. A photo passes through four stages, and each stage has its own ceiling and floor. The diagram at the top shows the first two.

1. At the sensor

This is the clipping that matters most, because it is permanent. If a photosite filled up during the exposure, the raw file holds the maximum value and nothing more. No software, raw or otherwise, can find detail that the sensor never measured. The range a sensor can hold between its ceiling and the noise at the bottom is its dynamic range, often 11 to 14 stops on a modern camera at its base ISO, and less at high ISO.

2. In the camera’s rendering

To make a JPEG, or the preview you see on the rear screen, the camera fits the raw data into 8 bits with a contrasty tone curve. That curve throws away the top and bottom of the raw range: in the diagram, the JPEG keeps about 8.5 of the raw file’s 12.5 stops. The brightest part of the raw, often somewhere around half a stop to a stop above where the JPEG turns white, survives in a raw file and is lost in a JPEG. This is why a raw file can “recover” a sky the JPEG shows as white, and why raw photos look different from the camera preview. It is also the reason to shoot raw when the light is hard: see RAW vs JPEG for the rest of that choice.

A plain sRGB encoding puts a middle gray of 18 percent at a value of about 118 out of 255, and pure white sits about 2.5 stops above it (log₂ of 1 / 0.18 is 2.47). So in a JPEG, anything more than about two and a half stops brighter than what your meter aimed for is at or near the ceiling. Picture styles with more contrast push that lower.

3. In editing

You can clip a photo that was perfectly exposed. Raising exposure, whites or contrast, dragging the white point in, or crushing the black point for punch all push tones past the ends of the scale. Sometimes that is the look you want; often it happens by accident while the photo just looks “brighter”. Editing in 16 bits and watching the clipping overlay while you work keeps it a choice. The tools themselves are covered in curves and levels.

4. On the way out

A wide-gamut edit holds colors that a small color space such as sRGB cannot. When you export, saturated colors outside the target space are pulled in, and the most saturated ones clip: a deep blue flower or a neon sign goes flat. Printers have even smaller gamuts than screens. Soft proofing, or a slight drop in saturation before exporting your photos, keeps detail in those colors.

Highlight clipping and shadow clipping are not mirror images

Most explanations treat the two ends as opposites. They behave differently, and that difference shapes how you expose.

Highlights hit a wall. The sensor ceiling is sharp. One third of a stop below it, a cloud has full texture; one third of a stop above it, the cloud is a flat patch. There is no warning band in the raw data itself.

Shadows sink into noise. At the bottom, a raw file rarely clips to a hard zero. Instead, the signal gets weaker until it is smaller than the sensor’s own noise, and the detail drowns. You can lift those shadows in editing and see something, but the further you lift, the grainier and blotchier they get. A hard, pure black in a finished photo usually comes from the rendering or the editor setting the black point, not from the sensor. That is why underexposure is mostly a noise problem and overexposure is mostly a clipping problem.

The top stop holds the most information. Raw values are linear, so each stop down halves the numbers. In a 14-bit file the brightest stop uses values 8,192 to 16,383, half of all the levels the file has; the next stop down uses 4,096 to 8,191; and the darkest stops get only a handful each. Exposing as bright as you can without clipping puts more of the picture into those rich upper stops and gives cleaner shadows. That idea has a name: expose to the right. Note that this is about noise and signal, not about running out of levels: bit depth explains why.

Put together: protect highlights first, because a clipped highlight is gone and draws the eye, while a dark shadow can usually be lifted with some noise as the price.

Why do lifted shadows turn grainy, and where does that grain come from? Noise explains it.

One channel can clip while the photo looks fine

Clipping does not have to be white. Each of the three color channels has its own ceiling, and a strongly colored subject fills one channel long before the others. A red flower in sun has a red channel near the top and green and blue far below it. Brighten it, and red hits the ceiling while green and blue keep rising.

The arithmetic is the whole story. A color is the ratio between its channels. While every channel doubles together, the ratios stay put and the color stays true. Once one channel is capped, the others gain on it: in the diagram, green goes from 28 percent of red to 44 percent. The flower drifts from deep red toward orange, gets paler, and loses the folds of its petals, because every petal now has the same red value. You see this in:

• red and magenta flowers, red cars and red clothing, which go flat and orange
• sunsets, where the red and orange of the sky near the sun turn yellow and then white
• blue sky near the horizon or behind a bright subject, which turns cyan as blue clips
• stage lighting and LEDs, where a pure blue or red light becomes a flat blob
• skin under warm light, which goes yellow-orange and waxy as red clips first

A single luminance histogram can miss all of this, because the overall brightness of that red flower is well below white. The RGB histogram shows it as one channel pressed against the right wall. Saturation explains why very saturated colors run out of room first, and RGB color covers how the three channels combine.

Want to read each channel’s graph on the back of your camera? How to read a histogram shows how a color can clip while the overall graph looks safe.

How to see clipping before and after you shoot

• Live histogram. On a mirrorless camera or phone you can watch the graph while you compose. A tall pile pressed against the right edge means highlight clipping; against the left edge, shadows going to black.
• Zebras. Many cameras can stripe the areas above a set brightness in live view. Set them near the top of the scale and they show what is about to clip before you press the shutter.
• Highlight warning. In playback, the clipped areas flash. These are the blinkies, and they point at the exact part of the frame that has gone.
• RGB histogram in playback. The only camera display that shows single-channel clipping. Worth turning on for flowers, sunsets and stage light.
• Clipping overlays in editing. Most raw editors can paint clipped highlights in one color and clipped shadows in another while you adjust. In many editors a single key toggles them, and holding Alt (Option on a Mac) while dragging a whites or blacks slider shows only the clipped pixels. Where the editor shows small clipping indicators above its histogram, they often take the color of the clipped channel and turn white when all three have clipped.

One caution runs through all the camera tools: they are built from the JPEG preview, not from the raw file. They warn a little early for raw shooters, by roughly the margin in the top diagram, and their timing changes with picture style, contrast and saturation settings. Treat a small blink as “close to the edge” and a large blink across your subject as “too far”.

Want to watch clipping appear on a live picture before you try it on a real scene? The camera simulator has a histogram and highlight stripes that appear as you overexpose.

Which clipping matters: a quick check

Almost every high contrast photo clips somewhere. The question is never “is anything clipped?” but “is anything I care about clipped?” Ask:

1. What is clipped? The sun, a bare bulb, glints on water or chrome, and the catchlights in an eye are meant to be white. These specular highlights are brighter than any diffuse surface and nobody expects texture in them.
2. Is it the subject or does it touch the subject? A white bird’s feathers, a wedding dress, a face lit by sun or the clouds that make a landscape need detail. A small patch of sky at the edge of a street scene usually does not.
3. Is it on purpose? A silhouette clips shadows to black by design, and a bright, airy high key photo may let a white background go. Clipping chosen for a look is a tool.
4. Is it a color? A clipped channel on the subject changes its color, which reads as wrong even when nothing is white.

Preventing clipping at the camera

When the check says something important will clip, work through these in order. Each costs a little more effort than the last.

1. Lower the exposure. Dial in negative exposure compensation in an auto mode, or a faster shutter or smaller aperture in manual, until the warning leaves the part that matters. Spot metering on the brightest area you want to keep, then opening up by about two stops, places it just under the JPEG ceiling; AE lock holds that reading while you recompose.
2. Shoot raw. It keeps the margin above the JPEG ceiling and all of the shadow range, so you can lift the dark areas later.
3. Bring the contrast down in the scene. Add light to the shadows with a reflector or fill flash, move the subject into open shade, or wait for softer light.
4. Hold back the sky with a filter. A graduated neutral density filter darkens only the top of the frame. How to use graduated ND filters covers placement and strength.
5. Take more than one frame. When the scene is simply wider than the sensor, bracketing gives you one frame for the highlights and one for the shadows, to merge or blend later.

Try it on a scene that is too wide for one frame: slide the exposure and watch the red edge appear at one end of the histogram or the other, then bracket and merge.

Scene still too wide for any single exposure? HDR photography explains when merging frames is worth it and how to keep it natural.

What highlight recovery can and cannot do

The highlights slider in a raw editor looks like magic the first time it brings back a sky. It is doing two things, and knowing them tells you when it will work.

It uses the hidden margin. The editor’s default rendering, like the camera’s JPEG, pushes the top of the raw range to white. Pulling highlights down brings that compressed top end back into view. If the raw data was below the sensor ceiling, the texture is there and comes back cleanly.

It rebuilds from surviving channels. Where one or two channels clipped but at least one did not, the editor estimates the missing values from the channel that still has detail. That restores texture, but the color is a guess, usually a neutral or slightly off tone. It works well on white clouds and poorly on a red flower.

It cannot invent data. Where all three channels hit the ceiling, there is nothing to rebuild from. Pulling the slider turns flat white into flat gray, sometimes with a color tint, and the edge where it meets real detail gets harder to hide. If the blown area is small, cloning or a careful local edit may hide it; if it is large, the frame is a lesson for next time.

Shadows are the reverse: there is usually something there, but lifting it lifts the noise with it. A moderate lift and some noise reduction is often fine; a four- or five-stop lift on a high ISO file rarely is. For selective work on one tonal range at a time, luminosity masks let you treat highlights and shadows separately.

Common mistakes and misconceptions about clipping

• “Any clipping ruins a photo.” Specular highlights, light sources and deliberate silhouettes are supposed to clip. Judge what is clipped, not whether the histogram touches the wall.
• “The highlights slider fixes blown skies.” Only when the raw data stayed under the sensor ceiling in at least one channel. Fully clipped is fully gone.
• “If the histogram is clear of the right edge, nothing is clipped.” A luminance histogram can hide a clipped red or blue channel. Check the RGB histogram on colorful subjects.
• “The camera’s warning is the raw file’s limit.” The warning comes from the JPEG preview, so raw files usually hold a little more. Learn your own camera’s margin (see below) rather than guessing.
• “Shadows clip just like highlights.” In raw, shadows mostly fade into noise. The hard black usually comes later, from the rendering or your own black point.
• “The rear screen looked fine.” Screen brightness and daylight glare make the preview image a poor judge. Use the histogram and blinkies, not the picture.
• Clipping in editing without noticing. Raising contrast, whites or saturation can clip a well exposed file. Keep the clipping overlay on while you edit.

Try this: find your camera’s real highlight margin

Ten minutes, any camera that shoots raw. You will learn how far past the blinkies your raw files still hold detail, which is worth more than any number on a website.

1. Find something white with fine texture in even light: a towel, a textured wall or a white shirt in open shade. Set the camera to raw plus JPEG, a picture style with normal contrast, and manual exposure.
2. Meter the white object so it sits at 0, then take a frame. Add one third of a stop and take another. Keep going in thirds until the blinkies cover most of the object, then take three more frames past that point.
3. Note which frame first showed blinkies on the camera.
4. Open the raw files in your editor. On each one, pull exposure down until the white is a light gray, and zoom to 100 percent. Find the brightest frame where the texture of the weave or surface still comes back in full, with no flat patches, and with the clipping overlay showing nothing on the object.
5. Count the thirds between the first blinking frame and that last good raw frame.

How to judge it: if you counted three thirds, your raw files hold about one stop past the camera’s warning at that picture style. If you counted zero, the warning is accurate and you should treat blinkies as the limit. Either way you now know how much to trust the warning, and you can repeat the test with a colored object, such as a red cloth, to see how much earlier a single channel clips.

Clipping FAQ

What does “clipped” mean in photography?

It means part of the photo went past the brightest or darkest value the camera or file could store, so it was recorded as one flat value with no detail. Clipped highlights look like flat white patches; clipped shadows look like solid black areas.

Is a blown-out highlight the same as clipping?

Yes. “Blown out” or “blown highlights” are everyday names for highlight clipping. “Crushed” or “blocked” shadows are the names for the dark end.

Can you fix clipped highlights in a JPEG?

Not really. The JPEG has already discarded the raw file’s extra margin, so a pixel at 255 has nothing behind it. You can darken it to gray or paint texture in by hand, but the original detail is not there.

Why does my sunset turn yellow or white near the sun?

The red channel clips first, then green, so the warm colors near the sun lose their ratios and slide toward yellow and then white. Lower the exposure until the RGB histogram shows red just off the right edge, or bracket for the sky.

Should I always expose so nothing clips?

No. Let the things that are meant to be bright clip, such as the sun, lamps and glints, and protect only the highlights that carry detail you want. Exposing so the sun keeps detail would turn the rest of most scenes nearly black.

Does shooting raw stop clipping?

It reduces clipping caused by the camera’s processing, because the raw file keeps a margin above the JPEG ceiling and the whole shadow range. It cannot stop clipping at the sensor: if the photosites filled, the raw file is clipped too.

Ready to test how well this has stuck? The dynamic range quiz asks about highlight recovery, shadow detail and when to bracket.