See: http://www.cambridgeincolour.com/tutorials/levels.htm

In the examples on that page, the original image doesn't have any pure white or pure black in it, so contrast can be added by changing the white and black points.

Its been my experience that many images that should/could have very bright/dark parts don't, and as such look better/more accurate after levels adjustment, but obviously there are images where this doesn't make sense (e.g. a picture of a seen that is mostly all greys, as an extreme example).

So the question is how does a camera decide where to put the scene's dark/light spots in the histogram? It could put them at either end of the histogram, which I'm sure it does in some cases, but in many cases this would be wrong, and vice-versa.


4 Answers 4


For the most part, it simply doesn't. The camera sets an exposure. At least in the simple case, it takes whatever amount of light is coming into the meter, and sets the exposure to make that into a fixed level somewhere in the vicinity of 18% grey. The more complex case is the multi-spot metering (goes by several names, but makes little real difference). Based on its measurements, it might decide to adjust the target exposure level to (say) 12% or 27% or whatever, but that's about it -- it's still only 1) measuring the light and picking some combination of shutter speed, aperture and (possibly) effective ISO to achieve the exposure it's picked.

While most digital cameras do some post-processing, it's not (at least normally) adjusted based on the content of an individual picture -- it simply has a tone curve for a particular setting, and adjusts tones based on the curve you've selected (high contrast, portrait, scenic, etc.)

Depending on the input you capture and the curve you select, that might be about the right contrast range, or it might be too little or too much. In theory, it could do quite a bit more, analyzing the histogram to decide on adjustments in roughly the same way as something like ACR does for its "Auto" exposure adjustment -- but at least in most digital cameras you don't get anything very similar to that.

  • 1
    \$\begingroup\$ This is correct. I want to emphasize that it works this way because most cameras do not adjust their response dynamically and once they have metered and a certain input level produces a certain output level, everything else falls into place. Some cameras can adjust their response though. Fuji calls this simple 'Auto Dynamic-Range', while Sony calls it Auto DRO, other names exist. In this case there is flexibility but it has a limited range. The response can generally accept up to +1 EV except for Fuji's system which can expand to +3 EV. \$\endgroup\$
    – Itai
    Nov 2, 2010 at 4:05

The answer to this is fairly complex, and is often dependent upon features and hardware specific to camera brands and models, as well as the camera exposure settings chosen by the user. To keep it simple, what the camera "sees" and what it decides to expose is dependent upon light metering. Modern cameras have sophisticated metering devices built into them that measure the light coming through the lens. Depending on how you have your camera configured, the camera will use the metered light value(s) to set aperture, shutter, and possibly ISO. When a scene is properly metered, a camera in fully automatic mode will usually choose correct exposure settings, however some scenes need more care and attention when metering to assist the camera in its choices.

Metering in most cameras is based on the ANSI standard value of "12% gray". This value is considered to be the "middle tone" between pure black and pure white in terms of luminance (light from a light source that is reflected off a scene or object in a scene). This means that the meter takes the average level of luminance of the area metered, and assumes the average is 12% gray. For scenes that cover a wide range of tones, from deep blacks through middle grays to bright highlights, this works quite well. For scenes that do not evenly span the range of tones, such as high-key or low-key scenes, the camera's meter may make an incorrect assumption about a scenes luminance, and measure 12% gray even if it should have metered a higher or lower value. Without careful metering with the camera, and use of the proper metering mode (more on this in a moment), such photographs often require black and/or white point selection during post-processing for correction.

Most DSLR cameras have a variety of metering modes. The default and most automatic is a form of evaluative metering, which meters a variety of areas in your scene and tries to apply an intelligent algorithm to arrive at a correct value. This often works great, but sometimes it does not work quite so well. Alternative modes include center-weighted, partial, and spot metering. These options measure progressively smaller areas, usually centered, although some cameras like Nikon allow spot metering around the currently selected auto-focus point. Spot metering is fairly precise, using only a very small percentage of the scene around the metered spot, to determine luminance. When using spot metering, it is best to point the camera at an area of the scene that is as close to "middle toned" as you can.

Not every metering mode works for every scene, and it is important to use the correct one. When manually setting exposure, it is often useful to use the camera's built-in meter in spot-metering mode to meter various parts of a scene to determine the real contrast (dynamic range) of the scene you are trying to photograph. This can be very helpful in helping you determine if you need filtration, or whether you need to adjust your lighting if you have artificially lit your scene. If you spend the time to learn how to use your camera meter in its various modes, you'll be more capable on a shoot or out in the field, and your scene contrast problems will eventually be a thing of the past.

Here are some useful articles on metering:

NOTE: Quite often, you may hear the value "18% gray" used as the luminance value that camera's meter at. Such a value is generally inaccurate if you wish to be precise, as 18% gray patch is generally considered to reflect half the light that reaches it. There is no direct correlation between a camera meters "12% gray luminance" and prints "18% gray reflectance", although I think generally speaking they can be considered roughly equivalent in their respective domains (i.e. One would expect a photo of a 12% gray card should print 18% gray, which when illuminated and photographed should be correctly metered at 12% gray again.) More details on this here.


The theoretical/philosophical answer would be that with each level setting you lose some information from your picture file.

Camera gives you pretty much fixed dynamic range, quantized (usually) over 3x256 levels. Would the range be incredibly wider, you would have the 'interesting' parts of the picture spanned over really small number of levels (the most obvious problem resulting from this is striping instead of smooth gradients). Would it be smaller (this is what you are asking for), you would lose some information from the edges of the interesting range, seriously limiting your ability to post-process the image.

So, to put it simply, treat the camera output as raw material, never as a ready-made product.


Part of the answer is that it is rooted in our preferences. Most people prefer images with a full range of tones that includes a small amount of full black and a small amount of full white. The original might not always be like this but most times we still prefer it to be adjusted to the full range of tones.

You ask 'how does the camera decide?' and there are some good answers above. But you should note that these methods are essentially rooted in film days when there was no other method.

Today we have the histogram and this changes the equation completely because it is based on all the information in the image while the metering methods are based on only partial information.

The metering method gives you a starting point so that you can get a histogram that is reasonably close to what you want. Then by examining the histogram you can quickly decide on the optimum exposure. Most often that is by adjusting the exposure to move the histogram up or down until it is nicely centered without too much pure black or pure white, but that is scene dependent.

In short this means you are taking control and can make more intelligent decisions than the camera.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.