> Someone claimed, in the discussion, that it depends on the color filter of
> the sensor. For CMOS sensors "...manufacturers broadened the light
> transmission of the sensor filter to let more light in but weaken the color
> differentiation."
That is even more true with Canon sensors than the others. They biased
towards light sensitivity than color sensitivity. In MOST cases, I
think this solution is better behaved as it maintains a more accurate
and usable gradient. But the problem comes in with color subtlety
allowing for saturation adjustment.
The analogy I like to use is comparing a hand-tinted B&W print with an
actual color photograph. The Canon images are much closer to the
hand-tinted print in theory and in practice. The image is actually a
B&W images with color applied on top. While this is technically true
with all digital camera images, the color information is weaker and
will use a mix-minus algorithm to strengthen the color information,
otherwise the bit-depth of the colors is too low. This is how the
human vision system also works. We actually cannot "see" red. We see a
color band that includes red. We also see green. (and blue). The human
vision system determines whether or not something is actually "red"
through the absence of green.
Do I see Orange? Yes
Do I see Green? No
Subject is Red.
Do I see Orange? Yes
Do I see Green? Yes
Subject is Orange or Yellow
Do I see Blue? No
Subject is Yellow.
Raw image conversion must use this same type of algorithm in order to
exaggerate the color bit depth. If each sensel (pixel) position on an
camera sensor is only 50% dense (color density - 100% would let in
zero light that isn't that color), then the bit depth of the color
information is cut in half. This color information has to be made up
somewhere else. Mix-minus is a way to recover some of that bit depth.
Regardless of how the color information is derived, during the raw
conversion process, it is generally applied after the brightness value
has been determined. Which is also why there are two greens to every
one blue and red. It has nothing to do with sensitivity, etc., it has
to do with the triangle intersection of three pixels. With the
exception of the Olympus CCD sensors (as well as several other cameras
using Kodak sensors), raw conversion uses THREE pixels, not FOUR! The
quickest way to determine if the converter has to use four is whether
there is a gray line at the intersection between two contrasting
colors. Anyway, I digress.......
When Phase One changed over from CCD to CMOS, they had to work closely
with the sensor fab people to completely customize the filtration
process as well as adjust the raw converter to correct for the
different technologies. It took years to get CMOS to work right for
Phase One.
AG Schnozz
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|