At 3:38 AM +0100 1/18/04, Listar wrote:
>Date: Sat, 17 Jan 2004 16:55:09 -0800 (PST)
>From: Tim Hughes <timhughes@xxxxxxxx>
>Subject: [OM] Re: Free OM-to-E-System adapter?!
>
>Winsor wrote : I don't think CCD vs. CMOS makes any difference as far lens
>performance is concerned.
>
> Winsor,
> There may be differences in the sense that the metalization layer and
>filter layers on the chip surface, may make the sensors deeper into the bottom
>of a "well"
>for one technology versus another. So the color fringing near sensor edge and
>the
>light drop off could theoretically be a bit worse for the technology with
>deeper wells.
>This is also true for different sensor area fill factors, although most
>sensors seem to have
>similar fill factors of the order of 50-60%.
This was my impression as well. For photographic applications, CCD versus CMOS
is a wash.
For astrophotography, CCDs still rule, but this is a price-is-no-object world.
>An interesting aside somebody made at a
>talk I attended, was that although lensed chips increase sensitivity, the on
>chip lenses do
>not work well for very fast lenses, causing increased light dropoff, so the
>super fast
>lenses have an effective slower maximum aperture than you might expect from
>marked
>F-stop. This is classic Murphy's law : When you really need the low
>noise/speed of your
>sensor and lens at low light levels, you don't get it. Obviously you can
>still use dof
>characteristics of the faster aperture.
It makes sense to me that fast lenses would suffer a loss, because the faster
the lens (the smaller the f/stop), the greater the included angle of the cone
of light converging on each pixel, so more and more of that cone of light falls
outside of the acceptance angle of the lensed pixel and is lost.
To put some numbers on it, the Kodak KAF-5105CE CCD ship used in the E1 has
lost about 1/3 stop at 12 degrees off axis (in the vertical direction). What
is the smallest f/stop (fastest lens) whose light cone fits within? F/stop is
simply the focal length divided by the exit pupil diameter, so the cone angle
from centerline is ArcTan(fl/(diameter/2)), which implies that the effective
f/stop is 1/(2 Tan(12 deg))= 1.86, call it f/1.9, which is pretty fast, so I
don't know that this effect will be that important in the E1.
I assume that there are other CCD chips with stronger pixel lenses, and thus
having more restricted acceptance angles, that will suffer when a fast lens is
used. Given that zoom lenses are commonly slower than prime lenses, this may
not be a problem in the intended application.
> The CMOS Foveon sensor should theoretically be much better in regards to
>using non-digital lenses as the wells are much shallower and the sensor area
>is at least 3
>imes larger per pixel, dramatically reducing the bottom of the well effects.
In an editorial appearing on page 59 of the 3 February 2004 issue of PC
Magazine, John C. Dvorak says that the Sigma SD10 with the Foveon X3 sensor
"produces the most stunning images of any digital camera I've used". Don't
know what else he has used, but given his position, it's probably a lot, even
if PC Mag isn't a photo magazine. Digital images are becoming a prime driver
of PC sales, so the computer mags are taking notice.
Joe Gwinn
The olympus mailinglist olympus@xxxxxxxxxx
To unsubscribe: mailto:olympus-request@xxxxxxxxxx?subject=unsubscribe
To contact the list admins: mailto:olympusadmins@xxxxxxxxxx?subject="Olympus
List Problem"
|