What is white or black "clipping", and how does one avoid it?

Question

I have been told that this photograph has "white clippings"?
What are they, and which light sources should be placed in what way to avoid that?

Answer 1

Clipping is a term that basically refers to data loss in a captured image. Another common description for this phenomenon is referring to part of an image as 'blown out'. The light streaming into your camera hits the pixels on your camera senor and is turned into a tiny amount of electric charge. Each element can only hold so much charge so when its full the highlight is fully saturated and when a whole area of pixels are saturated, any detail in that area is lost, referred to as clipped or blown. Similarly there has to be a sufficient amount of light hitting the pixel to generate any measurable charge and if there is not enough in an area of the sensor only a solid black is captured and any detail in that area is similarly lost. No amount of post processing can bring detail back in these areas because no detail at all has been captured.

With a film camera the same thing happens except that instead of the light being turned into an electric charge on a sensor it causes photosensitive molecules on the film to react. When they have all reacted in an area of the negative no detail can be captured in that area.

In compositions where there are both very bright and very dark areas such as your example it is very challenging to capture detail across the whole image. Techniques such as HDR were developed exactly for this scenario where several images at different exposures are combined to increase dynamic range. It is impossible to capture detail across the whole image with a single shot without employing some external lighting to illuminate the darker areas so you can lower the exposure an bring the lighter areas in. Professional photographers tend to make good use of reflectors to do this just to lift the shadows enough to get some depth.

Answer 2

The intensity, or brightness, of objects in the real world operate a lot differently than in a photograph. In reality, the brightness of an object, for all intents and purposes, is infinite. A powerful light bulb might look particularly bright, yet compared to the sun, its rather dim. The overall range of possible levels of light intensity is immense in the real world, ranging from dim starlight (say 0.0001 on a hypothetical scale) to sunlight (100,000,000 on the same hypothetical scale). This range of intensity is what we call dynamic range.

The human eye is capable of perceiving a limited dynamic range, and it cannot see both the dimness of starlight and the brilliance of sunlight at the same time...you can see one or the other. If your eyes are adjusted to see starlight, the sun and anything lit by it would effectively be "clipped" as far a your vision and perception are concerned. Conversely, if your eyes are adjusted to see the world illuminated by sunlight, the dimness of starlight would be well below the darkest parts of the world around you...effectively clipped into shadow. What is amazing about the eye, however, is its ability to adapt...the total dynamic range the eye is capable of functioning is extremely large...smaller than the total range of possible intensities, but far larger than common electronic devices like cameras and computer monitors.

Similarly, camera sensors and computer screens have an even more limited dynamic range than the human eye. Different than the eye, however, is the fact that digital devices must represent dynamic range as discrete values capable of being represented digitally. Digital devices are also limited in the total range they can represent...with total black usually being represented internally by the number zero, and total white being represented by some finite maximum such as 255 (8bit), 4096 (12bit), 16384 (14bit), or possibly as high as 65536 (16bit) on the most recent and top of the line cameras and monitors.

This range is considerably more limited than the possible range of light intensity in the real world, by a factor of over 1500 times. When exposing a photograph, one has to be aware of the limited dynamic range. Expose too long, and you risk capturing more light than is possible to represent in 8-16 bits of information...at which time, you clip any excess analog value to the maximum possible digital value. In the photograph you have posted, it appears that the collar of the womans top has been overexposed, resulting in it being clipped. Her hair also appears to be a bit under-exposed, and while its not possible to expose less than 0, it is possible to expose too little, such that the electronic noise of the sensor itself overwhelms any useful image data.

It is possible to determine before capturing a photo with a digital camera whether you might be clipping the whites (or highlights, as they are usually referred to), by using the histogram. The histogram is a simple diagram that plots how many of each tone (a level of intensity, ranging from zero to maximum) is present in the photograph. The histogram normally progresses from left to right (however some cameras are opposite), with the darkest tones to the left, midtones in the center, and highlights to the right. If you are over exposing, the right-most tones will be maxed out...reaching the top edge of the histogram. When you see such a histogram, adjust exposure downwards until the highlights are flat or just begin to rise near the right edge. It should be noted that if you expose for the highlights, you might lose proper exposure elsewhere. Some cameras contain built-in modes, such as Canon's Highlight Tone Priority, that will attempt to preserve highlights automatically without radically changing the rest of the image, which might be of some use.

Answer 3

Here is a histogram (Colors | Info | Histogram in GIMP) of that photo:

That peak at the far right implies clipping: in this case there were a lot of pixels that were too bright for the exposure settings and were all given the same (pure white) value. Generally you want to avoid a peak at either edge, since that often indicates the exposure isn't set right for the scene.

A way around this is to check the histogram display on your camera after you take each picture: if you see a peak at one end of the histogram, you are losing some information and you should adjust your exposure (note: there is a common piece of advice to "expose to the right," which is actually what you did here) to compensate. (If you see a peak at both ends (and your picture does have a small peak at the dark end), you're losing both light tones and dark tones; one solution is to use HDR techniques.)

What is dynamic range and how is it important in photography? explains the concept a little more.

Answer 4

White clipping is when a pixel (or area of pixels) not intended to be 100% white have been exposed just too bright and have clipped at 100% white (255, 255, 255). The shirt collar would be considered too clipped for comfort. That being said, if you were to look at the raw file for that photo (did you shoot it raw?), I would guess that most of the information is still in there and all it would take is a bit of an exposure adjustment to bring the detail back in.

Now, the real problem is when you clip the raw file's 100% white. This cannot be corrected for and usually looks fairly unpleasant.

Answer 5

A little bit of googling would probably answer your question.

Here's a link to a Wikipedia article on clipping.

From that article:

Bright areas due to overexposure are sometimes called blown-out highlights or flared highlights. In extreme cases, the clipped area may appear to have a noticeable border between the clipped and non-clipped area. The clipped area will typically be completely white, though in the case that only one color channel has clipped, it may represent itself as an area of distorted color, such as an area of sky that is greener or yellower than it should be.

Answer 6

Clipping is another term for saturation.

UPDATE: I am not referring here to the Color Saturation operator or variable (as in Photoshop or HSV), but rather to the mathematical/technical term. Sorry for the confusion.