Flat output A900 and D3X comparison

[David Kilpatrick]

Continuing the examination of the D3X and A900 files, now I have to the time (though not as much time on my hands as some, I guess) - by reducing contrast to -50 and zeroing brightness, using a Linear output curve then setting EV correction to place the highlight end correctly, and finally adjusting the Nikon to -10 Brightness to bring the gamma into line with the A900, I have two further comparison images:

A900:



http://www.pbase.com/davidkilpatrick/image/109078947

D3X:



http://www.pbase.com/davidkilpatrick/image/109078949

The captions on pBase give a fuller explanation. These are visually similar to gamma = 1.0 16-bit exports, the idea is to get every possible recorded step from the sensor into the file without any clipping. Some bright reflected highlights go beyond the sensor recording of either camera (or any camera) and you can see how these are handled on the miniuature guitar clock. Bits to look at: the top left hand edge of the bottle and the lettering on the shade card behind it; the shadow detail inside the duduk (pipe) visible through the fingerholes; the ivory handle of the nail file; the cap on the reed of the duduk.

My conclusion is much the same as before, that Nikon is applying a transfer function curve data before it is written (A to D transfer curve, or one in later processing) while Sony is either not applying this, or is using a flatter (lower gamma) curve. The second possibility, supported by Raw Developer's visible curve functions for the two cameras, is that ACR is applying a brighter function curve to the Nikon files by default - which would work, assuming the Nikon EI rating is putting the exposure 'to the right' accompanied by clipping the unused low-bit data.

Diagnosis: Sony should have released the A900 as an ISO 50-3200 model (100-1600 before expansion) which is essentially what Nikon did.

David

[Dr. Harout]

Thanks David, this one is much more convincing.