V1.02
Thom Hogan’s Complete Guide to the Nikon D300
Page 143
compression. Still, the size savings can be impressive: a
JPEG
basic
image can be less than 20% the size of a
JPEG fine
image. Yet despite that data reduction you’re not seeing an
obnoxious level of artifact build up.
Still, I’d say use
Optimal quality
and
JPEG fine
whenever
possible, as it clearly has small advantages over any of the
other renderings. Whatever you select as your “acceptable”
level of JPEG compression, start using better quality levels as
you get to ISO 1600, and bump up another level at ISO 3200.
JPEG Rendering
The D300 renders JPEG images a bit differently than earlier
Nikon DSLRs. While not publicly talked about, apparently
Nikon is using a NuCore JPEG engine to do the actual
rendering. What is known is that all manipulation of the
image is done using 16-bit data (early Nikon DSLRs used 8-bit
processing, later Nikon DSLRs used 12-bit processing). The
translation to 8-bit JPEG on a D300 is done only after all the
demosaicing, color manipulation, sharpening, and other
effects are first handled. In other words, the D300 takes the
12-bit raw data, renders that into a 16-bit data set, uses that
16-bit bit data set and camera settings to render a set of 16-bit
pixels, and then reduces those to 8 bits only at the point
where the actual JPEG encoding is performed.
While this doesn’t sound earth shattering, it does have an
impact on the images the D300 produces. When you apply
sharpening, tone curves, or color manipulation on 8-bit data,
you risk posterizing bits of the data. Posterization means that
some bit values just don’t exist. Too much posterization can
result in visible artifacts, usually “bunched up” or unnatural
tonal ramps. The problem is compounded if you take an 8-bit
