When Nikon released the D2X, I had doubts about the company's decision to stick with the 23.7x15.7mm sensor format. The 1.5X apparent focal length magnification factor didn’t bother me, but the use of smaller photosites (pixels) in the sensor worried me, especially given that it is a 12.4-megapixel camera.
The D2X sensor photosites are slightly smaller than 5.5 microns. Conventional wisdom says smaller pixels are more susceptible to electronic noise than larger ones, especially at high ISO-equivalence settings. Until recently, the predominant view was that a photosite needed to measure about 9 microns across to get a good clean signal.
High ISO equivalence in a digital camera is obtained by turning up the gain on the signal, which increases the noise-to-signal ratio. Artifacts from this electronic noise show up in the image as both graininess and as weaker performance in the shadow areas, where the lower light generates a weaker signal to start with. Like tuning in to a distant a radio station, turning up the volume increases the background hiss.
This noise can be at least partially processed out of an image either in-camera or in post-capture processing, and like all good digital cameras, the Nikon D2X includes an optional noise suppression feature for the ISO 400 to 800 range. There is an optional high setting, which is always active when the ISO sensitivity is set to Hi-1 or Hi-2 (ISO 1600 and 3200).
Using the camera's Noise reduction feature increases in-camera processing time and lowers image buffer capacity. After some experimenting and image evaluation, I decided to keep it off and use Noise Ninja from www.picturecode.com for shots taken at ISO 800 and higher. At very high ISO settings, shadow tones are not quite as smooth as I’d like them to be.