|
 Here's some hard truth for you: Proper exposure is no less important in digital photography than it is in shooting with film. In fact, digital capture is less forgiving in this respect than familiar color negative films. Digital SLR photographers need to adjust their exposure techniques to suit the inherent characteristics of the digital medium.
To understand why, and how, you need to know about the nature of digital capture. Most film exhibits a nonlinear sensitivity to light: The top and bottom of its tone curve, which represent light and dark tones respectively, start to "shoulder off" into a less severe, more gradual slope (see figure below). A digital camera's image sensor, on the other hand, responds to light in a one-to-one fashion, producing a "straight line" tone curve from lower left to top right.
Think of each pixel on the sensor as a bucket, and light as rainwater; once a bucket is full, any additional water (light) simply spills out, uncaptured. On the other hand, even when it isn't raining light, the pixel buckets are never completely empty: There is always a bit of water left at the bottom. In the digital world this is called the "noise floor"; it is there because all electronics generate a residual random signal even when not recording anything. The noise from your stereo when it is turned way up with nothing playing and the snow on your TV when you can't get a channel are examples of this.
What's more, as I explained in my column on JPEG vs. RAW in the last issue, an electronic image sensor records fully half of all the available tonal data in the brightest stop of exposure. The next-brightest stop captures half of what remains, and so on. Thus, a typical 12-bit digital SLR records a total of 4,096 brightness levels: 2,048 in the first stop; 1,024 in the second; 512 in the third; and so forth. By the time you get to the shadows there's little data left, so, given the ever-present noise floor, you'll want to put as much of that data as possible on higher reaches of the tone curve. In simpler terms, you bias your exposure toward the brightest values.
There is a catch. (Isn't there always?) If you're shooting in JPEG mode and bias your exposure too much toward the brightest values, you risk overexposing the image and losing highlight detail. (The bucket overflows when it's full.) Moreover, with just eight bits of tonal information, a JPEG's severely limited range of bright-nesses -- 256 levels versus the 4,096 that your D-SLR's image sensor can capture -- doesn't lend itself to major tonal adjustments after the fact, using software.
|
