3 okt 2008 kl. 16.02 skrev David Dyer-Bennet:
On Fri, October 3, 2008 08:26, karl shah-jenner wrote:
my understanding is that the name aside, digital cameras are still
analogue
video still cameras, with the analogue image digitized in camera,
subsequently they still behave much as a video camera does .. hence
the
effect shown in the original images Andy sent, and as experienced by
others
out there too including me.
Not all of them, not always. Some of them have mechanical shutters,
and
read out the sensor chips only when the shutter is closed -- hence it
doesn't matter what pattern they read it in. Others, especially
cheaper
and smaller ones, use "electronic shutters" meaning they read the chip
while it's being exposed, and hence have the potential for this sort
of
amusing aliasing.
I think the readout from CMOS vs. CCD chips may be very different,
too.
Michael Reichmann, on his Luminous Landscape site, has just published
a review of the video facility of the new Nikon D90: http://www.luminous-landscape.com/reviews/cameras/d90.shtml
Some way down, he discusses a phenomenon he calls "JelloCam" (well
worth reading in this context). He states that: "The cause is that
unlike CCDs, CMOS sensors do not record the entire image
simultaneously. Rather, they record from top to bottom, and so it's
possible for a rapidly moving object to appear in different places on
the same frame."
Usually, Michael is very well-read and reliable in his writings about
digital technique, so I assume he knows what he´s talking about here,
too.
In my previous post, I stated that those prop shots must be scanning
artifacts, which sems to be correct. But I also stated that there
would have to be simultaneous scanning of several parallel channels to
get it. Andy´s simple but ingenious experiment proves this to be
wrong. I eat my words...
Per Öfverbeck
http://ofverbeck.se
"In a world without walls or fences, who needs Windows or Gates"
