The comparisons between the sensor formats is quite interesting. There are
all sorts of magical properties which are being applied to the
medium-format digital cameras which have multiple causes and
misconceptions. Medium Format Digital has greater dynamic range. Uh, yeah,
right. Some talking heads will yap about multiple stops. Uh huh. Try one
stop, maybe two, over a typical 35mm digital under IDEAL conditions. Sorry,
that's not enough to explain the differences in the images.
Medium Format lenses are designed differently with different contrast
curves and intents. Even among the 35mm manufacturers there are great tonal
differences between lenses of different manufacturers. What is assumed to
be an inherent advantage of the larger sensor is actually related to the
lens designs, not the sensor designs. We saw the same thing back in the
film days. A cropped image from a Hasselblad looked different than the same
image from a Canon 35mm camera even when the subject size on the film was
identical and the print size was identical. When subject-on-film size and
subject-on-print size were identical you eliminate the variable of film
size and are looking strictly at the properties of the lenses (and
vibration of the systems). Those same "oooh, look how wonderful the
medium-format images look" characteristics in the digital age were also
talked about long ago in the film age.
Of the sensor designs, there has been a tremendous difference in NATIVE
tonal and spectral curves between CMOS and CCD. That's changing as both
technologies are getting closer and closer to each other, but earlier
generations of CMOS had substantial "bit-bending" to make the colors line
up. This "bit-bending" did reduce the overall dynamic range OR the
microcontrast in particular colors. To identify this and quantify it
requires careful controlled testing with color test charts. Taking a
picture out in the wilds is not usable evidence. (although over a period of
time with multiple images, certain trends will emerge). This kind of
careful controlled testing is pretty much unseen today. DPREVIEW is the
only testing company that comes close. DXO is on some planet of their own
making. Any reviewing company that comes up with their own definitions of
"ISO" is going to be suspect in this regard, because the tests are easily
exploitable once the testing methodology is identified.
I believe that what is seen in the on-screen, highly-shrunk, images is the
emergence of microcontrast. We run into this to a great degree in
black-and-white photography. It's easy to make a system generate a mostly
linear response curve from 0-100% brightness. It's another to make the
region from 30-70% look correct with a natural gradient. For example, if I
do a B&W photo of a foggy scene, I can stretch the contrast to give myself
a black and a white point within the image, but then the midtones (of which
made up the majority of the original scene) are non-existent. In an 8-bit
image editing world, you'll get steps in the tones. With a technology that
retains a smooth tonal gradient within this highly stretched region, you'll
get a more pleasing microcontrast--the ability to distinguish slight
differences between similar tones instead of having the tones move to the
nearest brightness level. When this happens, we get tonal steps. In the
chemical B&W darkroom, this is a problem we run into with multi-grade
papers--There are two cross-over points where the tonal curves have bumps
in them which cause the final print to have points where some tones are
brought up in brightness and others are depressed. Ilford Delta 100 has a
specific notch like this between Zones II and III where both II and III are
merged together to the same brightness level. Single-grade papers, unlike
the multi-grade papers achieve a near perfect tonal response curve without
the bumps which results in a tremendous range of microcontrast across the
entire brightness spectrum. In the film world, the quality of the
microcontrast is governed by the distribution of random sized particles of
silver. In the digital world, randomness from one sensel to the next is
important.
The Olympus E-1 is one camera that has a distinct "medium format digital
look" because of the randomness in noise and sensitivity levels in the
sensels. Even at ISO 100, if you photograph a blank scene or have areas of
great smoothness (bokeh blurring), you'll see what appears to be noise in
the image. This noise is present even in the brightest areas. This
"dithering" is nearly identical to the dithering applied to audio in
professional audio production. It causes a randomness in each byte that is
similar to the randomness that occurs in film by the varying sizes of the
silver particles. Provided that we don't enlarge the image too much, the
E-1 image will retain a greater sense of color depth than the average bear.
The medium format digital cameras tend to have similar traits, but with
greater numbers of pixels so they hold up to the enlargement far better.
(on a few magnitudes better). Upon downsizing for web display, this may
reveal itself as one file having a larger color gamut over another.
Since Brian mentioned the E-3, it has been my experience with the E-3 that
even though I can get it to match colors with the E-1, the apparent
microcontrast and color gamut appears different. I VERY MUCH see this with
the DMC-L1. The older Panasonic sensors produced images that can be made to
match images from other sensors in tonal range and response, but the images
are distinct from different cameras. Under some lighting conditions one or
the other cameras will crash-and-burn in tonal smoothness. The E-1 will
bury the propeller in the ground when it comes to smooth blue skies--the
blue sensitivity range is highly stunted (by almost two stops) and
stretching the contrast of a sky will result in banding. Until I did
controlled testing with color charts and raw file analysis software did I
understand what was happening. When it comes to reds and greens, the E-1
absolutely rules, but once you start mixing the blue sensels in, the
microcontrast gets a bit raspy. The DMC-L1 is almost the opposite--the
blues are better, but the reds crunch up.
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/
|