Wayne wrote:
> We've had this discussion before....
Yup. And I've smoked this weed before too. :)
> As a circuit designer, an A/D does not do processing, it is linear. I can see
> post processing of the A/D information, or have a non-linear gain in front of
> it.
That is absolutely true. In a simplified form, all the
analog-to-digital converter does is record voltage levels from a
sensing device of some form. What I'm talking about is what goes on
when photons are converted to electrons and while the electrons are
being moved around, but before the electrons are counted.
Just as film can be presensitized, a sensor can be presensitized with
electrons to the point right below the threshold where the first
photon results in a measurable electron. Well, that's fine and dandy
and pretty much how every camera sensor works today. But Olympus/Kodak
took this an additional step with the E-1 where additional random
voltage changes were made to provide dithering to not only smooth out,
but shape the response curves with a gamma adjustment. It also appears
that Kodak attempted to soften the clip point (rather unsuccessfully,
I might add) with the three color sensels clipping at different
points. In a straight conversion, the shoulder isn't bad, but if you
apply highlight-recovery in the computer, you can see how horrible the
E-1 really is up there. Given the relative lack of horsepower
available in the cameras at the time, it was probably faster to adjust
things in the analog realm instead of through computation.
> The main difference in color reproduction is the color filter array and its
> ability to separate colors. Sharper filter separates the colors more. If the
> color filter is less sharp, there is no way to really recover the lost
> information. Dxo Mark has camera sensor color response and typically green
> bleeds into red and blue. No E-1 information though. When green filter
> overlaps/bleeds into red, what do you get? And how does the camera interpret
> it?
This is an area I've harped a little bit (little bit???) with the
Canon. The Canon images tend to be a B&W image with color tinting
applied, whereas, on the opposite extreme, you have the E-1 that is a
blob-o-color with luminance channel applied. Both can be valid, but
the way to get there is quite different. Canon's method more closely
resembles that of the human eye. The human eye sees luminance, blue,
green, orange. Red is a calculated interpretation as a result of a
lack of green. Honestly, I think in the world of LED artificial
lighting, it's probably better with the less coarse filters. In the
natural world with smooth, wide-band "white light", I think the
sharper cut filters are better.
> The E-1 also has a thick AA in front of the sensor which can probably do more
> NIR filtering. Not to mention that it renders dust less of an issue by
> keeping it further from the sensor.
This is probably one of the most brilliant aspects to the thick
optical stack. However, it does mess up the focus point slightly with
non-ZD lenses. The actual focal plane is not the sensor surface, but
somewhere in that glass stack.
> I am happy to sacrifice some color fidelity for higher dynamic range, lower
> noise, more convenience/performance. My color IQ is rather low anyway.
I'd say "it depends". For the 90% rule, I completely agree with you.
However, it's that other 10% why I like the Ford Edsel of DSLRs.
> After I'm done processing images, are they really what I actually saw?
> Generally the light was flatter and colors less saturated than what I create
> in post, but probably no where near as much as many photos I see online these
> days.
Good question. My thinking is what IS REALITY is not MY REALITY. My
personal perception of the visible world isn't always in line with
what the visible world really is like. I just happen to really like
the way the E-1 sees the world as it comes closest to the way I see
the world. With every other camera, I can create this alternative
reality, but not exactly the same way. Just because Lightroom has a
Saturation slider doesn't mean that you're going to get E-1 results
from a 6D.
This just goes to show just how insane I am.
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/
|