I wasn't trying to disagree with Chuck on the diffraction limits
thingy. The hard facts are that F16 is going to stink on the E-1 as
you are far enough past the theoretical limit that it doesn't matter
if it is F8, F11 or F2 that the calculators come up with.
Another aspect of this discussion which isn't taken into account very
well. The diffraction limits for digital imaging assume that each
sensing pixel position is a stand-alone affair on the sensor. It is
not. In fact, this whole calculation is based on pixels being isolated
from each other, but the very nature of the bayer array utilizes
"nearest neighbors".
With most cameras, the algorithm is a variation of the following:
Take one green, one red and one blue pixel and create an RGB value for
an implied pixel at that intersection. The Olympus is a little
different and harkens back to some early sensor designs which use four
pixels for the algorithm. In fact, the Olympus (specifically the Kodak
sensors, but even the Panasonic sensors show this characteristic) use
two different types of green sensors. If you don't include both greens
as part of the algorithm, you'll get lots of artifacts in the image
and any variation of the color yellow will turn UGLY on you.
OK, so with the E-1, as we average four pixels together to get the RGB
value, but that would result in a 50% drop in effective resolution.
Not quite. Most algorithms will step forward one pixel at a time to
recalculate. But with the Olympus E-1 files, most algorithms that do a
single pixel step will not map the greens correctly and will produce
artifacts. To get around that, they usually do a two-pixel step. Those
algorithms DO result in a 50% drop in effective resolution. The
in-camera JPEG engine of the E-1 does this very thing. However, from
my own testing it appears the higher-res engine that was found in
Olympus Studio actually did a per-pixel step. However, where most
algorithms for most sensors are tri-pixel based, an E-1 sensor needs
the algorithm to calculate the intersection of four pixels.
In essence, the E-1, with the dual-color greens, is one of the most
complex sensors to convert the images from. Your choice is to have
improved resolution (but if you do that, make sure that NO channel is
within one stop of clipping), or to have color fidelity without
artifacts.
Back to the diffraction-limits. The E-1 image files are the spongiest
files around. Getting anything super-duper sharp is an impossibility.
There is no such thing as pixel-sharp, no matter how many times in the
past I've referenced it. Canon cameras can produce pixel-sharp images,
Olympus cameras generally can't. And the E-1, with the combination of
the hyperactive AA filter and the 4-by pixel-matrix takes it into
another whole dimension of futility.
Therefore, whatever the calculator says is the diffraction limits of
the E-1 are probably a whole stop off and even at that you are really
hard pressed to see where it makes much of a difference. F16? Yes, you
are well past the edge of the cliff, but depending on the focal-length
and lens, the diffraction limits are pretty much theoretical.
As always, test your own equipment. Know your own equipment. Don't
take Chuck's word, my word, or some calculator's word for gospel. No
two lenses are the same. No two camera models are the same. And no two
RAW converter algorithms are the same. In-camera JPEG will look
different than an Olympus Studio converted file which looks different
than an ACR converted file.
AG
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|