----- Original Message -----
From: "Peifer, William [OCDUS]" <[EMAIL PROTECTED]>

(snip)

> I'm curious where this whole idea of CCD sensors requiring (or
> preferring) perpendicular rays originated.  I'm pretty convinced that
it
> must have originated because somewhere along the line, something got
> taken
> out of context, and a fundamentally incorrect idea grew from there.

When the MZ-D got canned it was reported (and commented on in PDML) that
the full format imaging chip was experiencing problems with its off-axis
image capture, caused by the physical construction of the pixels
overlayed by the colour matrix.  I also commented on this in the thread
"OT: DOF and format size".

> From
> the standpoint of the underlying physics, Tom is absolutely right --
the
> purpose of a lens is to bring an image to critical focus at the focal
plane,
> and the nature of the sensor (film, CCD, CMOS, or other) isn't
particularly
> relevant.  After all, if all the light rays strike the sensor
> perpendicularly, then they are necessarily parallel and thus cannot
form an
> image at the focal plane!

No one suggested that the image forming rays must only be parallel in a
CCD equipped digicam.  Rob Studdert's observation was that a symmetrical
design of WA lens might create problems because its image forming rays
would be very (very, very very!) far from perpendicular at the corners
of a full format chip.  If the back focus of the lens is long, ie by
using retrofocus designs for the WA lenses, then the difference between
the angle of the image forming rays and perpendicular will be smaller.
If the exit pupil of the lens is large comparative to the format size
then that difference will be reduced too.

> I suspect that this perpendicular-ray story -- dare I say "legend"? --
may
> have originated from a misinterpretation of the characteristic
behavior of
> CCD sensors.  We all know that in single-chip color CCD sensors, some
of the
> pixels are sensitive to red, others to green, and still others to
blue.  For
> the case of color cameras with single CCD sensors, color sensitivity
is
> imparted to a particular pixel by incorporating a microscopic optic --
a
> lenslet and filter -- in front of that pixel, which I believe is
> accomplished as part of the manufacturing process for the sensor chip.
I
> can imagine that the numerical aperture of this microscopic optic may
not be
> terribly large, and it might very well constrain the field of view of
its
> corresponding pixel.  Maybe someone that knows more about chip fab can
> comment on this.

(snip)

It's about geometry.  For the sake of illustration I'll assume the
individual pixels are circular (they may in fact be any shape their
maker prefers).  A perpendicular light ray as would be found on the
optical axis of a lens will be able to expose the entire circle.  A
light ray travelling to the corner of the chip will find a pixel that
appears to be oval-shaped because of its angle of incidence, and that
oval shape will expose less area to the light rays and gather less
photons as a consequence.  The amount by which a CCD is inferior to film
in this regard might be small, but because some lenses already give
marginal illumination to the format corners the inferiority could be
enough to make a lens useless.  The Philips chip was reported to have
its pixels slightly recessed behind the colour matrix as a result of
which the light rays were significantly obstructed from exposing the
off-axis pixels.

The spiel that came with my scanner (Plustek) remarked that CCDs in
scanner duty give great DOF and are well suited to 3D objects unlike
CMOS arrays that only record sharply what is directly in contact with
the platen.  As scanner arrays are not masked by a color matrix this is
a pretty good indication that CCD pixels do indeed have an narrow field
of view and hence a long DOF.  Whether or not CCDs have long DOF in
camera duty is another question, but don't expect an informed answer
from me :)

Regards,
Anthony Farr
-
This message is from the Pentax-Discuss Mail List.  To unsubscribe,
go to http://www.pdml.net and follow the directions. Don't forget to
visit the Pentax Users' Gallery at http://pug.komkon.org .

Reply via email to