Exposure: Over, Under, or Neither?

Over the past few years I have heard and read various arguments for how to expose digital images. Over-expose? Under-expose? Neither? I want to offer a few thoughts about balancing these arguments in light of each photographic situation.

For starters, if it needs to be said, virtually everyone who can shoot "RAW" images should be doing so. A camera-raw image holds much more information than a compressed JPG file which the camera can make by throwing away information it decides will not be needed, in a split second based on predetermined criteria chosen by the camera manufacturer.

It might be a perfectly fine JPG file, but it's a reader's digest version of the full RAW file.

And secondly, a very common photographic challenge is to somehow capture and store (either on a piece of film or a digital sensor and file) the large range of dark and bright information which our eyes can so easily take in.

With photographs, often a sacrifice must be made, to preserve the details in the brightest areas of an image at the expense of the details in the shadows; or vice versa. The "dynamic range" of a human eye is huge - the range of brighter and darker exposure values which can be taken in as part of a scene.

The dynamic range of color negative film is about 8-10 exposure stops. The dynamic range of color slide film (which Kodak just discontinued - Kodachrome, the longest and best regarded slide film) is about 6 exposure stops. Because slides are projected as-is, they need to be exposed correctly.

The dynamic range of a digital sensor today is 5-12 exposure stops depending on the camera. The higher dynamic range is only found in higher end DSLRs costing several thousand dollars. This is rapidly changing of course, as the search for the digital sensor holy grail continues.

That should be enough background. Regarding digital exposure, there are three approaches: over-expose by about 1-2 stops; under-expose by about 1-2 stops; or do neither -- seek an "accurate" exposure for the subject that neither favors the highlights nor the shadows. The game plan, in all cases, is to try to recover useful image information (captured by the sensor and stored in the RAW file) in the darkest and the brightest areas of an image.

1) The over-expose strategy. This is sometimes called "Expose to the right," which refers to the histogram: choosing to expose images so that there's more information crowding the brighter (right) side of the histogram (which all DSLRs and some point-n-shoot digitals provide on the LCD display) and there is little or no "clipping" of blacks and shadow detail on the left side of the histogram. The argument here is that it's easier to recover highlight details out of the nuanced exposure information from a RAW file than it is to do so with details in the shadows.

2) The under-expose strategy. Not everyone agrees with the "expose to the right" approach. I just listened to a podcast where a respected photographer was advocating exposing for good preservation of highlight details even if this means the shadows are clipped and end up crowding the left side of the histogram, with the expectation that these can be recovered relatively more easily from the RAW file than is the case with highlights. Basically, approaches 1 and 2 are opposites. Opinions differ, and people can back their opinions with anecdotal experience.

3) The "neither" strategy. When shooting slide film, this was the only choice. Shooting even 1/2 stop over or under would result in an image on slide film (projected on a screen) that looks incorrectly exposed. And even though digital sensors and raw files allow for a range of choices to be made after the image is captured, some would advocate that the best approach is to accurately expose the overall image on average, leaving room to pull details from either the shadows and the highlights (or both) back into an edited image for the screen or for print.

My argument with all the above is that it depends. Specifically, it depends on at least three things, and we're assuming for this discussion that the image is being recorded to camera raw, not just to a JPG file. It depends on the subject, the camera, and the intent for the image.

First, what is the subject matter? Does the image being captured have a huge dynamic range of 15-20 stops of information, such as a sunlit marketplace with trees and shiny roofs and faces peering out from doorways and windows? Or is it a soft lit portrait of one of those faces surrounded by the darkness of the interior? Is it a macro of a flower? Is it a bride in a white dress, with African skin tones, and surrounded by a wedding party with varying skin tones and some of them wearing black tuxes?

In each case, how the image should best be exposed depends on the subject and the light which makes it visible to the camera. There is no simplistic, universal approach like "expose to the right" or "expose for the highlights" or "just expose it accurately!" :-)

Second, what camera is recording the image, and what kind of digital raw file does its sensor create? Is it capable of some of the highest dynamic range that today's best sensors can achieve? Or is it a 4 year old DSLR which makes a very nice image but may have only 7-8 stops of dynamic range on a good day?

Third, what is the intent for the image? If it is for family prints and shared Flickr sets, or a print ad campaign, or a fine art gallery show, or wedding enlargements, or a coffee table book -- each of these intended uses for the images might inform the best arguments for how best to expose the images. There's no one golden rule of exposure that best suits the many ways images might be used.

One rule that does apply to almost all images, whatever the exposure, is that post-production is an important part of making the image. A careful study, on a calibrated monitor, of the shadows and highlights, and an evaluation of what information can be recovered, and which information should be favored when we can't "have it all." Because we can't. Photographs don't yet store the information our eyes could see when we saw the scene. The challenge is to make our best effort at capturing the best essense of a scene and representing it later.

How do you expose digital images? Over? Under? Neither? I'm interested to hear your comments.