Many photographers using digital cameras make it a policy to "expose to the right". This has nothing to do with political orientation, rather it refers to exposing so that the exposure histogram is as far to the right as possible without clipping. Actually some clipping is fine as long as you understand and agree with what parts of the image is being clipped. If fact, you should read all the way to the end of this article and discover why you may actually want to be pushing the cameras clip limits in practice.
A common approach to exposure is to expose to get something that "looks good" on the LCD on the back of the camera. This is flawed in at least two ways. First of all, camera LCD's are notorious for being a very bad way to evaluate exposure. Shots that look great on the camera LCD can look badly exposed (not to mention unsharp) once they are evaluated on a good monitor. Even more fundamental is the fact that in most cases the LCD preview is derived from an 8 bit JPEG reduction done by the camera.
Photographers who are "ETTR shooters", if we can use such a phrase, are also going to be capturing and later working from RAW captures. Their aim is to get the most and the best data into the raw files, and they will make it look good later. In many cases an ETTR image will look overexposed on the preview - trust the histogram!
I have read two somewhat different justifications for ETTR. The first one states that a camera sensor is basically a linear device, whereas "stops" are essentially logarithmic. This means that half of the linear codes generated by the analog to digital converter used to read out the sensor fit into the upper stop of the sensors range. Not exposing to the right means that lots of resolution is being wasted. This is the argument put forth in the now famous essay on the luminous landscape site:
The second argument goes along the lines of more light. Exposing to the right means exposing more (letting in more light). More light means more signal, which means high signal to noise ratios. This is the argument put forward by Jeff Schewe. See pages 41-45 in the image sharpening book by Fraser and Schewe or the essay in the following link:
Oddly enough this has been a topic that has stirred up a lot of discussion and controversy. John Shaw even chimes in on the topic with some very useful observations.
First of all, my hat is off to him once again for his, "get your camera and do some experiments" approach (his books recommended doing this to find out what ISO setting you should really be using for your specific camera with different makes of film).
Second of all, he points out (in agreement with Schewe's focus on noise) that pushing this hard to the right is particularly important when using high ISO settings where noise is expected to be an issue (he mentions ISO 1600).
He recommends doing a test to find out how much "headroom" there is from when your camera warns you that highlights are being clipped and when they actually are clipped in the RAW file. He says that with his Nikon D800E, he has 1.3 stops of headroom! More than you would ever have imagined!! In other words, with the camera holding this 1.3 stops in reserve, if you obey its warnings attempting to avoid highlight clipping, you are already throwing away that top stop with half of the sensor encodings that the luminous landscape post was warning you about!
Tom's Photography Info / [email protected]