Archived posting to the Leica Users Group, 2003/09/01

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

Subject: [Leica] Digital back for Leica M, sequel, precision and so on
From: Grégoire Vandenschrick <gregoire.vds@skynet.be>
Date: Mon, 1 Sep 2003 21:16:12 +0200
References: <BCEKKGNGDPMOIPMEJONBCECICKAA.phong@doan-ltd.com>

I see that others are now discussing the pro and cons of a digital back for
the Leica M.

I will not enter the fight that oppose I will call them the "purists" in one
hand, wich beleives only negs or dias and so on can deliver a good work, nor
I will prophetically announce the dead of the Leica M (and the film) for
tomorrow.

I gave you a few days ago my vision of what could be the digital leica M
(see archives) and had some very interesting answers.
I have gone on, and searched about CCD's and Leica lenses, and think more
about technical problems and practical solutions.

Here are my results so far:

The MAIN PROBLEM (the argument cited by Leica, will also not discuss if what
Leica says is correct or not, not my business) is, of course the angular
incidence of the light in the corner.
But not only, there is also another problem, which is the angle by wich the
maximum aperture is seen (the biggest in the center!).
this is true that with a smaller CCD, some brands achieved to have a rather
good picture, BUT, you cannot imagine to have a CCD smaller than 24*36mm in
a Leica M, because of the way you build the picture, through the different
frames in the viewfinder. If a smaller CCD than 35mm film is used, this
would say that every leica should have its viewfinder changed in some way to
have the new picture collimated seen! Nonesense.

So, we are here with a first technical problem, is there a full frame CCD
(24mm x 36mm) with a good resolution, and a good angular tolerance?

Take for example the Kodak KAI-11000CM (full frame 24x36 11Mpixel). everyone
can go to

http://www.kodak.com/global/en/digital/ccd/products/interline/KAI-11000CM/indexKAI-11000CM.jhtml?id=0.1.8.4.19&lc=en

have a look and download the different PDF. Read it, or read me: the angular
tolerance is so: if we say the light received normally (90°) is 100%, if we
fix that we don't want to have more than 10% of light falloff, the angles
are +-10° horizontally, and +-20° vertically. First stop to discuss this.
Why there is more angular tolerance vertically than horizontally. Quite
simple: the pixel is 9x9µm ok, but the photosite is 9x4.5µm, vertically
placed. The rest of the pixel is composed by two VCCD of 4.5x4.5µm, used to
take the electron charge from the photosite and pass it to the following
line (lower). So, due to the INTERLINE design of the CCD, the angular
tolerance is twice better vertically than horizontally, quite peculiar for a
lanscape CCD. So, first idea, why not make the same CCD, but INTERCOLUMN, or
let's say, not 4008 columns x 2672 lines, but rather 2672 columns x 4008
lines. Same clock timings, same size of pixel etc, but with +-20°
horizontally, and +-10° vertically.

Second idea, I gave to Solms, but still didn't have any answer, maybe some
of you could be rich implementing it, I don't currently have the know-how to
devellop it:

Someone talked about the microlenses which are placed before the pixel.
These are not used to compensate for the angular incidence of light, but to
compensate for the smaller size of the photosite. the photosite, even
regardless of the presence of VCCD's in the pixel, is not using all the
surface on which it sits. So, to compensate this, a "stupid" condensator,
like the ones before the TTL cell or ligh cell in your M6/M6TTL/M7/MP, is
build just on the photosite (at the very vertical of it, FOR ALL THE
PIXELS). And this is where one could work a little bit the things:

Breath, ok, seat, go:

so, imagine we have this KAI-11000CM rotated by 90°, we know that we have a
"pyramid" of light usable of +-10° vertically (parallel to the 24mm border)
and +-20° horizobntally (parallel to the 36mm border). You can also think
this pyramid as the image of the photosite projected by the condensator, or
microlens (or lenslet) array. I don't know precisely how these microlenses
are done, but they look very precise and well centered, so, the way they are
build is precise, and so we could build them in another pattern!!! Let's
imagine we minutely shift a microlens in front of the photosite, parallel to
the CCD, and in the direction of the picture center. the light pyramid has
changed, it is no more +-10°, but maybe +12°-8°. And, We don't need to
have -8° (light coming from a field larger than the film surface
(practically no leica lenses present this incidence, -1° full open for the
75mm summilux, -3° for the Summarit 85mm. So we could displace the microlens
in order to have a cover of -1° + 19° vertically. And this is true also in
the horizontal plane, where we could have, just by decentering the microlens
of each pixel, a light pyramid of -1° + 39°!! Of course the shift should be
exactly axipete, and different for each microlens, nothing in the center,
maximum in the corner.

I think one way to build microlens array is to attack a polymer with an acid
that can have access to the polymer only by round holes in a mask, which is
designed with another polymer hardened by UV, or punctured by laser, I don't
know. The acid is attacking the polymer from the hole, the chemical action
progressing with a sphere front from the hole. once the depth recquired is
attained (and so the curve of the lenses), the mask is removed, the acid
rinced, and this piece is used as a mold for the real microlens array. Ok,
so everything is a question of where to do the holes in the mask with my
laser.

Everybody could calculate this pattern, in order to have practically an even
light on all the CCD. the only problem is that you play only with this
pyramid of light, and according to my measure, if you want to bend this
pyramid in order to accomodate for the wide angle lenses, there is some
135mm lenses that are less well covered (but still), but I think everybody
would agree with this choice.

Ok, this is enough for this time, advices welcome, critics too, insults,
brevet proposal, industrial joint and other job opportunities greatly
welcome ;-)

I have two or three other trucks full of these thinkings if you want about
the digital Leica M, let me know, but please, do not answer me that the
angle of the light in the corner is too big for etc, I know this!!

Best

Gregoire Vandenschrick


- ---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.512 / Virus Database: 309 - Release Date: 21/08/2003

- --
To unsubscribe, see http://mejac.palo-alto.ca.us/leica-users/unsub.html

In reply to: Message from "Phong" <phong@doan-ltd.com> (RE: [Leica] Digital Leica M)