Monday, March 3, 2008
Turbo'd Image Sensors
At the heart of it all, the tiniest technology makes every picture possible
ABOVE: The sensor is one of several microchips in a modern D-SLR. This high-horsepower Olympus sensor relies upon a similarly high-horsepower AF system to be able to capture the image at its fullest potential.
The Matter Of Size
Though it seems bigger is always better in pixels and sensors, it's time to learn the exceptions that prove the rule. Some camera manufacturers—such as Olympus, with its Four Thirds sensor, and Nikon, with its pro-quality APS-sized CCD sensors—have found that a smaller sensor actually has advantages over big and expensive chips. In the first place, because digital cameras aren't simply replacing film cameras but allowing for the total redesign of a camera system, manufacturers can build smaller, lighter, less expensive cameras around sensors that meet those same requirements.
Because more and bigger pixels mean more densely packed sensors—unless the size of the sensor also increases (and even then, a large sensor still can be filled with pixels)—they can be more susceptible to noise. The advantage of a big pixel is better color and higher dynamic range (because of a bigger, clearer source in relation to the background noise from the charged chip), but the downside is more heat. This necessitates space for dissipation, thereby increasing the chances for moiré and aliasing. Everything is a trade-off, particularly at the microscopic pixel level.
Pixels are measured in microns, or millionths of a meter. The periods in this paragraph are 50 or 100 times larger than a typical 2- to 10-micron photosite. Whether they're large or small, the tighter they're packed, the greater the chance of another digital downer—blooming.
Blooming also could be called sensor spillage because it happens when a pixel well overflows with charge and spills out, contaminating the surrounding pixels. No problem with actual buckets and water, but in sensors it's a big deal. That electronic overflow means contaminated pixels in overexposed areas where photosites overflow—affecting everything from color to contrast and general usability of a pixel. Fewer usable pixels make for a less than ideal picture. Much like gutters on a roof, anti-blooming drains are incorporated to whisk away overflowing electrons and keep them from contaminating nearby pixels.
Speaking of neighboring pixels, it's impossible for most pixels to work independent of their counterparts on a chip—particularly in terms of color. Because pixels only read luminance, effectively seeing the world in black-and-white, they team up to interpret the actual colors of a scene they record. Sensor-makers create this connection by applying color filter arrays to the surface of the sensor.
The Bayer pattern is predominant among digital cameras, and it works quite simply. Imagine a checkerboard pattern once again, but this time imagine that one quarter of the squares are red, one quarter are blue and half of them are green. The rows are laid out in a green-red-green-blue pattern that accommodates the human eye's extra sensitivity to green light (particularly as it relates to sharpness). Each filter allows only its color of light to pass through to the pixel, accurately rendering that color on a given pixel. The sensor's pixels then work together, interpolating the color between pixels to determine how each one relates to adjacent pixels and, ultimately, to the full-color reproduction of the scene.
Page 3 of 4