The main issue in making good exposures in high contrast scenes is learning the difference between how your eye “sees” and handles contrast and how the sensor “sees” and records brightness values. Contrast is defined as the difference between the brightest and darkest areas in a scene. In photography the areas that define a usable contrast range are those in which you can see and record detail and tonal values; the compositional decisions often involve how you treat those brightness areas that fall outside this range.
For example, if you photograph a white car in bright light you would want texture and tonal value in the car body and details and perhaps even in the tire tread. But you might not care about the details in the asphalt that sits in the shadow of the car. Or, if you’re taking a portrait in bright light you’ll want good skin tones values in your subject but may not care about information (details) in the shadow he or she casts. When we talk about a usable contrast range we are talking about those areas that you want to record and not those that may also be in the scene but that can fall into tone without detail, like a deep shadow. We can call this usable range of values the "significant" tones, with the brightest in which you want texture the significant highlight and the darkest in which you want detail the significant shadow.
If you take an exposure reading of just the significant highlight you are placing that highlight on the middle of the recording scale—-in essence, you are telling the exposure system that you want the highlight to record darker than it appears in the scene.
And, if you take a reading of just the significant shadow area (like in the image shown here) you will be recording it as brighter than it appears in the scene. This throws off the balance of brightness values in recordings where there are both bright and dark values. If you make a reading of and record the darker areas alone it will cause the brighter areas to “burn up” and become overexposed, just like the side of the building here.
If you want to make a quick test of how making readings from the lighter or darker parts of the scene affects your results, set up a bracketing sequence at +/-2 EV and take three pictures of a brightly lit scene with shadows and highlights. One exposure may average the two values, one will expose for the highlights and one for the shadows. You’ll see how making exposures for just a certain part of the brightness scale affects the other areas.
Photo and text copyright George Schaub 2010. Exposure here was read from the shadow areas. The result in a high contrast scene such as this is fairly substantial overexposure of the highlights, never a good thing in digital (or film) photography.
