I thought about how to approach this test a lot. From what I had read the extra color depth of
shooting in 14 bit color versus 12 bit color should show up in the shadows, and highlights in particular.
There would be a larger palete to represent all those details. I now fear that was wrong, but no matter...
here is what I did. I shot two dimmly lit objects that would contain a full range of luminosity...hoping
against hope that the added gamut would reveal the promised bonus of the added color depth found
in 14 bit images.
Exhibit A: An Egg
Exhibit B: A Golf Ball with dark turtle on top
I have looked at the 12 bit to 14 bit raw files, and those I ran through NX to 16 bit tifs and can not tell the
difference. I think I understand the difference in color bit depth, but I suspect the human eye can not distinguish
the difference. In a typical jpg file there are potentially 16,777,216 colors. Of the three luminosity channels (RGB)
there are 256 shades of red, 256 shades of blue, and 256 shades of green. That is 8 bits per color, or 256 x 256
x 256 = 16,777,216 colors. When a camera is shooting in 12 bits per color the three luminosity values contain
4096 shades of each primary color. 4096 shades of red, 4096 shades of blue, and 4096 shades of green, or 4096
x 4096 x 4096 = 4,398,046,511,104 potential colors. (that is right - over 4 trillion colors). When we raise the bit
depth to 14 bits per color - instead of 4096 colors per primary color - it goes to 16,384 shades of color per each
primary luminosity channel. This staggering number of potential colors is expressed like so - 16,384 x 16,384
x 16,384 = 281,474,976,710,656. or over 281 trillion colors.
The next issue is how many colors can a typical monitor display? Well. I can tell you that it certainly is not
that many colors.Neither CRT, or LCD monitors can display even 12 bits per channel, though some CRT
monitors may come close. Next we have the color space that file is shot in, and the color space the image
is display in. Adobe 98 has a broader gamut than srgb. When I looked for an absolute number of colors to
express this gamut - I can not find a reliable number to express how many colors are contained in each color
space. I did find an interesting letter online which asserts that Adobe 98 colorspace has about 51% of the
gamut of LAB color, and srgb had about 36% of the gamut of LAB color. (or 50% less than Adobe 98)
So, the colors our cameras shoot are being reinterpreted into a color space that contains fewer
colors than the info they were shot in. Then there is the working color space (if adobe 98) contains
about 2 million colors, and then the monitor may represent those colors using even few colors.
SRGB would start with even fewer colors than Adobe 98. The LCD that I use is rated to display
the SRGB gamut, and I would think almost any CRT is capable of doing the same. But, most
consumer grade LCD's are not capable of displaying that many colors. Include all Apple imacs,
and laptops - same is true of all LCD's used in the PC world. Unless, you purchase a very high
end LCD like Eizo brand, or other highend calibrated LCD monitor.
Comparison-3
What are you going on about? The short version is this. I can't see a difference between
the 12 bit and 14 bit Nikon files. It may be there, but my equipment isn't able to display it
in a fashion that makes it apparent to me. Think about it this way. Imagine 4 trillion widgets.....
now imagine 281 trillion widgets. I doubt anyone can imagine even 1 trillion of anything. Thus,
the human eye isn't able to distinquish that many colors, the software can't use a colorspace
with that wide of a gamut, and the display isn't able to display that many colors. So, does it
matter? I have to assume since they are making it a feature that it does matter to someone -
somewhere. Another way to look at this is that I'm completely lost, and don't know what to look
for in this test. If you can correct me, please do - I'm just a photographer.
EDIT***I may have made a mistake in believing a web site on the numbers of colors. 2 to
the 12th power is 68 billion, not 4.2 trillion, and 14 bit is 4.3 trillion not 281 trillion. However that
doesn't change the basic point that the hardware and software reduce the pallet of colors to the
same number - which is a fraction of the number recorded by the camera, and makes it very
difficult (if no t impossible) to see the difference. I did have one person suggest shooting a blue
sky, and I can see some merit in that idea. Perhaps a subject with a reduced range of color, and
luminosity is precisely the type of test to conduct. I will report back when it is complete.
