Source: The Luminous Landscape
Written By Michael Reichmann
You may not realize it, but your digital camera doesn’t give you every pixel that it records. In most cases this is because the manufacturer masks off pixels at the very edges of the frame. This is done for a couple of technical reasons.
1. Many of the image processing algorithms (de-mosaicing, noise reduction, sharpening) that camera firmware (or raw conversion software) apply to raw sensor data are not “point operations”, but are instead “neighborhood operations”. When computing the output value of a given pixel, the algorithm needs to know not only the input value of that pixel, but also the input values of the pixels in the local neighborhood. The pixels at the very edges of the recorded image don’t have a complete local neighborhood, so the algorithms need make some guess as to their value, which means that output results for these edge pixels will be not quite as accurate as the pixels a few pixels away from the edges. By trimming away these edge pixels, the camera manufacturers guarantee that all the output pixels are of maximum quality.
2. Sometimes image sensors don’t have aspect ratios that exactly match the standard aspect ratios (3:2, 4:3, 16:9, etc.). In these cases, there are often different numbers of horizontal and vertical pixels hidden, to cause the final output image to have an exact aspect ratio. When some users print digital camera images, they often don’t bother to crop them, so having the camera output images with standard aspect ratios that exactly match standard paper dimensions is an advantage.
Occasionally, your framing of a shot may be such that you need those hidden pixels. What to do?
Also, there are cameras appearing, such as the Panasonic LX1, and its sister camera the Leica D-Lux 2, which have 16:9 format sensors, but which can also be set to shoot 4:3 or 3:2 aspect ratio pictures. The camera actually records the sensor’s full area regardless of how the camera’s selector switch was set. But if you decide that you’d rather have the full width frame there’s no way to recover those missing pixels. Or is there?