Early examination of the Df and D5300 is producing some interesting low-level differences that will need some more examination to determine whether they are truly meaningful. To wit:
- D5300 — This is the first Nikon DSLR where the black level has not been centered on 0. Instead, black is defined at something around 600 (in 14-bit values). This has the net effect of placing black high enough in the bit values so that noise errors on each side of black now form a natural bell curve. All previous Nikon DSLRs truncated black at zero, so only half the noise errors were encoded in the data, which has some minor consequences for noise reduction routines that try to “fix” noise in areas of black.
- Df — This is not performing the same as the sensor as the D4. In the ISO values up through about 1600 the noise levels are lower, which seems to indicate that read noise must have improved in the sensor (that was the gating component of the dynamic range of the D4 at lower ISO levels, much like the D3, only not quite as bad).
Both sensors seem to still have a bit of deep shadow banding. Not enough that it worries me in any way, but those of you thinking you’re going to recover six stops of underexposure in the shadows should be ready for a surprise.
So Nikon is continuing to tweak the unseen-to-user aspects of the sensors it is using. Whether this has any real impact on images will have to wait until I’ve completed my testing.