[Leica] M9 - the end of the line for Leica?

[lrzeitlin at gmail.com (Lawrence Zeitlin)]

 Perhaps Dante Stella is correct in his assertion that the M9 may be the end
of the M line for Leica. Indeed it may be the end of the line for digital
cameras that emulate film cameras. As photographers who grew up using film
age and leave the active camera market, camera design may turn in a
different direction. We have already seen it happen in video cameras. Modern
camcorders have ilittle resemblance, either in form or function, to the the
old Bell and Howell or Bolex. Here are some thoughts on camera evolution,
probably brought on by too much single malt scotch while wearing my Tilly
hat. Perhaps we sometimes forget that it is the image, not the lens or the
camera, that is all important. If you are not interested in photographic
technology, read no further. Just go out and take pictures.

The future of digital cameras, even for Leica should the firm survive, is
software minimization of optical defects. Why not? This is nature's way. The
human eye, for example, has a primitive optical system, basically an F3.5,
25mm FL non-achromat doublet. The sharp image circle is about 3 mm in the
center of the fovea giving a field of view of 3 degrees in a single
fixation. The retina's "film speed" in daylight is roughly equivalent to ISO
800 with a central resolving power, under perfect conditions, of
approximately 68 l/mm. Looked at objectively, the raw image quality is about
the same as a Box Brownie. All those lovely, crisp, wide angle images you
perceive are constructed by software processing in the brain.

Here is a partial list of what goes on. The projected image is encoded,
focus is corrected, edges of objects are enhanced, colors are assigned to
various portions of the image depending on which cells in the retina are
activated, small image portions are stitched together as a function of
eyeball position to form a whole percept, and an illusion of depth is
created by the disparity of images from each eyeball. A pseudo image is
created for blank spots (blind spot) in the retina. In addition, geometric
shapes are corrected so that they accord with experience. Objects viewed at
a distance are made to appear larger. Colors constancy is maintained despite
changes in the viewing light. And the same image enhancement techniques
could be applied to tomorrow's digital cameras.

Microcomputer technology has reached the point where lens defects can be
corrected in software better than in glass. The defects that are most easily
fixed are those of geometric representation. Pincushioning and barrel
distortion corrections are almost trivial. Intensity fall off at the edge of
a wide angle image can be dealt with fairly easily provided sufficient
information exists to let the software boost or subdue a portion of the
image. Coma, astigmatism, and field curvature are harder to fix but all have
been done in special purpose packages. Most difficult, at least with a
planar sensor are color refractive aberrations.

The fact that most geometric aberrations can be corrected in software frees
up lens designers. They can concentrate on creating faster multifocal
lenses, or superior performance lenses with fewer elements. This is not a
new idea. Image defects have been corrected by other means for decades.
Extreme wide angle lenses (Hypergon, etc.) often used a variable density
filter, darker in the center than the edges, to equalize image density
across the frame. Some cameras, both cheap and very expensive, used curved
film planes to compensate for curvature of field. The Kodak Brownie, the
Minox, and large astronomical telescopes all used this trick. Even Leica
uses offset micro lenses to change the angle of light rays at the edge of
the field.

So here is what I envision for a future generation of digital cameras.  At
the final stage of manufacture, a lens, fitted to its camera body, is
focused on a diagnostic target. The image from the camera's sensor is
compared with a theoretically perfect image of the target. Pincushioning,
barrel distortion, and image fall off are measured, corrected and the
correction factors logged. If the lens is of variable focus design,
corrections are logged for each focal length. Areas of poorer resolution are
determined and local sharpening is employed to provide uniform apparent
quality over the field of view. A lookup table with all the corrections is
burned on a microchip and incorporated into the lens mount.

When the lens is mounted to the camera, the camera's microcomputer notes the
corrections necessary to get a perfect image with that particular lens and
adjusts its image processing to suit. In addition, dead pixels in the sensor
are mapped (hopefully very few) and a fill in algorithm is used to provide a
seamless image. Each lens that can be fitted to the camera carries its own
lookup table for a "perfect image."

Will it be costly? Maybe at first, but the magic of electronics is an ever
decreasing price curve. Once lens data chips and adaptive camera
microcomputers are mass produced, and automatic lens calibration systems are
developed, the cost will almost certainly be lower than conventional cameras
of equivalent performance. Faster microcomputers will make the new digital
cameras as responsive as the old film cameras. After all, it is much cheaper
to make a digital chip than a precision lens element. Did I mention that I
bought a half dozen full function scientific digital calculators for $1 each
at the Dollar store?

Nature doesn't depend on perfect optics to provide a perfect image. Why
should Leica?

Of course all this may never come to pass. Both the Mayans and Hollywood
predict that the world will come to an end in 2012. My ordinary Leica will
be good enough to take pictures of the last few minutes. Maybe alien space
travelers will see them.

Larry Z