Nikon D300 Complete Manual page 156

V1.02
Nikon has long differed from Canon in how "black" is placed
in the bit storage of raw files. This reason gets quite technical
and is generally only of interest to those using their camera to
shoot astrophotography (e.g. through a telescope). I'll cut to
the chase: essentially, Canon places the black value higher in
the bit storage and allows the "noise" in the black signal to be
fully recorded. Nikon places the center of the black signal at
0, essentially clipping some of the noise. This really doesn't
make any tangible difference unless you're trying to resolve
very low magnitude stars using image stacks.
Raw file compression has an effect on the other end of the
data (the highlights). Nikon now supports two types of NEF
compression: Lossless compressed and Compressed.
Obviously, the lossless option has no impact on the data: you
get back the original data when the file is uncompressed.
However the second compression option is what Nikon refers
to as "visually lossless." It does lose and modify data (more on
that in a bit in the section labeled "Compressed NEF").
But I'm getting ahead of the discussion here. Let's back up for
a moment. Data coming off the ADC portion of the sensor is a
set of either 12-bit or 14-bit raw information, essentially the
photon count for each photosite. For raw files, this
information simply needs to be saved into a file along with
the camera setting information (EXIF data) and an image
thumbnail. The image thumbnail is generated by the EXPEED
imaging ASIC while the raw data is in the memory buffer,
then the file is written to the card. The format used is Nikon's
derivative of the TIFF-EP standard. In other words, a Nikon
NEF is a TIFF file with Nikon-defined tags pointing to various
.TIF
special information, including the raw data. Instead of as
.NEF
the extension, is used instead.
This special TIFF (NEF file) has three primary data repositories:
the raw image data, the EXIF data for the camera settings, and
a JPEG thumbnail created from the raw data by the imaging
ASIC. If set to 12-bit, the raw image data is packed as two 12-
bit values in every three bytes (three bytes = 24 bits). If set to
Thom Hogan's Complete Guide to the Nikon D300 Page 156