Archived posting to the Leica Users Group, 2001/07/26

[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]

Subject: Re: [Leica] digital leica - cheap lenses plus software = zero distortion etc.?
From: "Mxsmanic" <mxsmanic@hotmail.com>
Date: Thu, 26 Jul 2001 12:46:01 +0200
References: <Pine.3.89.10107260313.B1795-0100000@post.cis.smu.edu>

Robert Monaghan writes:

> won't it be possible for a digital camera
> (Leica?) to use a map of the lens distortion,
> various aberrations, and other defects at
> various f/stops and focusing points, and then
> remove these distortions thru software?

It is possible as long as no image information is actually lost on the way to
the CCD.  In other words, if the CCD receives image information that is merely
distorted, but not erased, it can reconstruct the original image, but if data is
actually missing, reconstruction is not possible.

For example, with a precise model of the lens behavior and all data concerning
the exposure, it would be possible to correct for chromatic aberration, because
the image data is there, even though it has moved around a bit.  However, if
part of the image is significantly blurred, image data is actually lost, not
just transposed, and so even with a precise model of lens behavior, it may not
be possible to reconstruct the original image without defects.

It is interesting to note that the human brain does this sort of processing.
The human eye has a very simple lens that shows a significant amount of
chromatic aberration.  However, the brain knows the exact amount of aberration
and corrects for it digitally during the process of perception, so that we do
not see color fringes in our visual field, even though they are certainly
present on the retina.  The brain undoes the aberrations so that the image looks
exactly as it should.

At the same time, we cannot correct for a blurred image.  The image detail just
isn't there anymore, in any form, and so no amount of processing and no amount
of lens data can help.  This is true for both the brain and for a digital
camera.

> I presume this data could be generated for each
> lens during mfgering, and perhaps loaded into a
> ROM chip in the lens, and corresponding settings
> read off from the lens during use (e.g., f/stop,
> distance)?

All true, but at what price?  This sort of processing requires major computing
power, plus extra hardware for sensors and indicators, plus personalized
attention during manufacture.  It might actually be cheaper to just mount a
better lens.  In practice, though, I expect that nothing will be done at all,
since most users don't complain about lens defects.

I expect that the Leica digicams will be just as cheap and defective as all
other digicams aimed at the same markets and at the same price.  I'm pretty
convinced that all Matsushita wants is Leica's know-how in lens design, and it
is paying cash for this, in the form of a "partnership agreement" in which the
main contribution of Leica to the cameras will be a little red dot stuck on the
front.  Nobody is going to buy a cheap consumer digicam with a real Leica lens
in it, because it would cost too much.  Additionally, Leica itself could never
build such lenses in the required quantities.  Leica will probably just send
trade secrets to Japan, and get large checks in return.  When Matsushita has all
of Leica's secrets, it will dissolve the partnership.

> what lens faults  could NOT be factored out, and why?

Those that produce a permanent loss of image data, which would include anything
that creates ambiguity that cannot be resolved.  Many combined aberrations, lack
of contrast in the glass, etc., might fall into this category.

> Will this permit low cost future optics on a digital
> (Leica) camera to achieve much better performance at
> a lower price point?

No.  Either you must pay for better glass, or you must pay for a mountain of
technology trying to compensate for poor glass, but both paths are probably
about equally expensive.  In any case, you won't get better images without
paying more money.

> What qualities will current glass and film based
> Leicas have that will be superior to such a digital
> camera software based optimization process?

A superior lens is always preferable to a poor lens with software corrections,
particularly since software cannot correct everything.

But I really don't expect to see true Leica glass on any of these digicams.  It
will just be a Panasonic digicam with a Leica name on it, and a marketing
campaign designed to give the impression that the digicams are somehow better
because they contain some sort of pseudo-Leica glass.

In the meantime, Matsushita will use the real optical secrets for other
projects.

In reply to: Message from Robert Monaghan <rmonagha@post.cis.smu.edu> ([Leica] digital leica - cheap lenses plus software = zero distortion etc.?)