V1.02
Thom Hogan’s Complete Guide to the Nikon D300
Page 72
this—seems to be the D300’s weakest; the green and red
responses seem to be stronger and less prone to error.
Noise
Noise refers to pixel data values in your image that are
different from what a “perfect sensor” would produce.
For example, in a “perfect sensor” three adjacent pixels from
an evenly exposed gray card might be rendered with 8-bit
RGB values of 128,128,128. Most digital sensors aren’t close
to being that perfect (and note that there’s rounding going on
somewhere to get to an 8-bit value for JPEG images, which
slightly exaggerates noise), so you might have one pixel that’s
128,129,128, another that’s 128,128,129, and a third that’s
128,128,128. As noise increases, the divergence of those
values would increase. For example, if the proper value is
128,128,128, then a value of 122,124,127 is clearly “noisier”
(and less accurate) than one of 128,126,128.
Here’s a full size section and a
blowup of a small piece of a
D300 image taken at the
3200
ISO value and intentionally
underexposed by more than a
stop to present a worst case
situation (we’ll look at a similar
image again when I discuss ISO
later in the eBook). This is
about a 600x600 section from
the full image, and you’re
looking at it in 100% view.
