There are two primary methods of determining exposure. Understanding them and when to apply them will make it possible for you to more accurately exposure your files.
The most used method for most photographers is to base the exposure on the highest important values in the image, ignoring specular highlights. This follows the concept of ETTR or "Expose To The Right" placing the brightest values to the far right on the histogram without clipping those values (overexposure). This is based on how linear digital capture works in your camera. This makes the best use of the full dynamic range of the camera and minimizes noise generation in the file. It is excellent for landscape photography and many other types of photography where overall exposure of the scene is the primary concept.
This practice can be overly simplistic as image content can exceed the dynamic range of the camera. In some high contrast circumstances the highest values like the sky, and important lower values can be further apart than desired for optimal exposure of both areas. In these circumstances it may be necessary to make more than one exposure, exposing once for the best high value area, and again for the lower value area. This will provide better source material for post processing but will require more complex processing techniques. One option is using HDR (high dynamic range) software, but that often results in less than the best image quality. Another option is to blend the two exposures using masking which is a bit more complex but not unrealistically difficult and can be mastered with a little practice. The blending technique is my preferred solution to the problem.
The most important part of producing good exposures is learning to use the histogram to help you understand what your exposure decision is providing you. There is no such thing as a perfect histogram as it is simply a graphic representation of the values captured. How the values lies within the histogram is what is important, especially how the values lie in relation to the left and right end. Exposures that result in the histogram hitting the wall on either end are not desireable, but can be informative in determining exposure settings.
The camera histogram can be a bit deceptive as it is based on the camera processed jpg and not the actual raw file. Lowering the sharpening and contrast settings in the picture style section of the camera menu will help. It also helps to train yourself to not expect the preview screen image to accurately represent the finished image. The histogram information is far more important.
The second method is basing the exposure on the primary subject in the scene without consideration for the overall values in the capture. This is best used for portrait and wedding photography or other images where the background may be secondary to the subject. The primary subject of any photograph should be where good exposure lies, but in this method the primary subject and the background surrounding the subject may vary in values.
The zone system can be applied to digital photography if you test your camera to know the dynamic range of the sensor and where certain exposure values will fall in the histogram. This is a bit more complex than the average shooter is willing to take on but simply having a few target values in mind for highlights and shadows is often sufficient.
Exposure is a relationship between the various tonal values in the image. The differences in reflectivity of areas of the scene is what creates the image and sets the mood of the photo. If the differences are great the image is said to have a high contrast and may require special consideration such as exposure bracketing, HDR or exposure blending to reign in the contrast. A subject with less inherent contrast will have a flat appearance coming out of the camera and may require expansion of the differences in tonal levels to look like you want it to.
It is important to understand that setting highlight exposure should be the primary consideration. Shadows fall in relation to the highlights and in digital capture underexposed shadows are an invitation to noise in the image. If the shadows are too dark with an approrpriate highlight exposure the contrast of the scene exceeds the dynamic range of the camera.
This image was taken on the side of the road in New Mexico. The lighting was fairly soft as a result of the cloud cover. Despite some high values in the sky area, the vast majority of the foreground was limited to the bottom half of the exposure scale. If you roll the mouse over the image you will see the original color capture before Photoshop adjustments. You will see how the wood structure is not rendered well and the sky contrast is rather flat as well.
Initial exposure was based on the high values in the sky, placing them as high as possible without clipping. An overall exposure based on the camera’s meter would have blown out the sky. I find that determining exposure based on the highest values works best for me. Once that is done the shadow areas are checked relative to the highlights.
I find that my camera typically wants to underexposure most average scenes by as much as a full stop, so I check the highlights and the overall histogram carefully after each capture. In addition I use spot metering to check specific areas of the image and "place" the values where I want them to be. This is an old technique known as the zone system.
In the zone system you use the meter to determine the exposure value of an area of the image. The meter tells you what the exposure should be if you want the area to be middle gray. You then modify the exposure in order to move the value up or down and expose according to the desired level. In the case of highlights you increase the exposure by about two stops depending on exactly how bright you want the area to record. That sets your exposure and the shadow values are recorded relative to the determined highlight exposure.
Setting the exposure based on the highlights seems to be to be the best option as clipped highlights and blank white areas are not what I want in my images. Second to that is the shadow area exposure which will be rendered in relation to the highlights. All that remains then is a quick check of the histogram to determine if the scene falls within the desired range and unimportant areas are not clipped at the bottom of the exposure. Losing highlight detail is not an option for me, while losing shadow detail is less important depending on the image but is easily remedied by a second exposure if needed.
The histogram confirmed that the scene was within the dynamic range of the camera, meaning the exposure I established for the sky did not force the shadows into clipping at the lower end. There was a substantial amount of foreground detail available as the left end of the histogram did not “hit the wall”. This gave me a good exposure for post processing with a single raw capture. The hump at the left of the big dip in the histogram represents the foreground information and the hump at the right represents the sky area.
In post processing there was contrast added to the sky, which showed values from slightly below middle gray to the high 240s. There was also contrast added to the foreground, and in particular to the wood structure. In both cases the corrections were made using masks to limit the corrections to specific areas. Brightening the wood by raising the contrast in curves significantly increased the foreground interest and made the image come to life. The color version was interesting, but the image to me needed to be black and white. That conversion was pretty easy since the primary dynamic changes had been done in the color version, but final tweaks to the wood structure needed to be applied to the black and white version to get the real impact where it needed to be.