Monday, October 8, 2007
Exposing For RAW
There are special considerations to take into account if you're shooting RAW and you want to be sure that you're getting a proper exposure
When viewing the normalized “overexposed” images and the “normal” exposed images at 100 percent or higher, the ETTR technique always produced superior-quality images, but the degree was highly dependent on ISO. Progressively higher ISO set-tings resulted in better quality, yet in all cases, the differences of ETTR were visible. As the math suggests, less noise and often better shadow detail can be seen.
I still had to deal with the in-camera meter, which measures the reflected, not incident light, then come up with some compensation factor if I didn't want to resort to using an external incident meter, which is highly recommended. Reflective meters are calibrated to “see” or measure re-flectance, depending on make, of 12.5 percent or 18 percent gray. Aim such a meter at a white dog in snow and you end up with a gray dog. Point the meter at a black cat on a pile of coal and you end up with a gray cat.
Photographers still need to understand the limitations of a reflective meter based on what it's currently measuring and adjust accordingly. For example, an old analog film trick was to take a meter reading on your hand illuminated by the light you'd be shooting in, then open up one stop. Modern cameras have a number of metering modes that can measure multiple areas in an image and attempt to provide better exposures than a spot meter described above, yet the fundamental concepts are the same. Having produced a series of test exposures using an incident meter, I hoped to find the actual ISO sensitivity of the camera sensor based on the ETTR technique.
ETTR: Is It Worth It?
The practical implications of ETTR are to produce superior image data, especially with respect to noise in the shadows. The math, which suggests that ETTR produces better data, is undeniable, but does an ETTR technique have practical advantages?
For high-ISO shooting, it's clear from the test examples that noise is greatly reduced; however, shooting in this mode effectively provides a much lower initial ISO setting, so this technique might not be that useful. The noise-reduction advantages are less noticeable at lower ISO settings, and the disadvantage to this technique is the appearance of images displayed on the camera's LCD.
Getting acceptable-looking LCD previews and ideal exposure don't mix, not until the camera manufacturers radically alter the way they build the preview based on the RAW data being captured. The histogram provided is equally less useful because it's gamma encoded based on the camera-generated JPEG settings. Clipping indicators could be useful, assuming you test your camera system to see what clips visually on the LCD in comparison to what clips in your RAW converter. Exposing to avoid highlight clipping, at the expense of more shadow noise, is the lesser of two evils. Remember, the worst possible situation is blowing out subtle highlight data you hope to render in your image. However, the ETTR technique illustrates that underexposure causes excessive noise in shadows at all ISO settings, so this too should be avoided.