On Wednesday, August 23, 2006 at 17:01,
Chuck Norcutt <olympus@xxxxxxxxxx> wrote:
> I don't see how. The way digital image stabilization works is to allow
> the image to float across the surface of a sensor somewhat larger than
> the image area and capture the image area wherever it happens to be.
> How you tell where the image is supposed to be I haven't a clue. Maybe
> accelerometers?
>
> Next has to do with the sensitivity of the sensor. As the image is
> skittering across the sensor surface eventually the shutter button is
> hit and it's time to capture. Is the sensor fast enough to do that?
>
> Let's use some hypothetical numbers. Assume a 9.3 MP E-3 sensor with
> 2700 x 3600 pixels. (nice round numbers, I just made it up). Using the
> 14-54 lens at 54 mm the image covers 19.1 degrees horizontally. If the
> camera is rocking horizontally at a rate of 10 degrees/second (another
> number I just made up... but it seem reasonable) then the image is
> skittering across the sensor at a rate of (10/19.1) x 3600 = 1884
> pixels/second or 1.884 pixels per millisecond. Does this mean that
> stabilization only works at very high shutter speeds? Seems that that
> problem has already been solved by none other than high shutter speeds.
Why would stabilisation be limited to one pixel or two?
There are two ways I see it could work, for any shutter speed.
1. By taking many pictures (slices) within the shutter interval, comparing
them sequentially, and
creating a picture file with most movement removed, adding up the exposures
in pixels that
minimise differences between successive frames.
2. Using a an accurate motion/angle sensor with a measure of the zoom setting
and, again using
slices, calculating which pixel in each slice corresponds to a place on the
subject, then
creating the picture using the added-up exposures.
or, of course, both assisting. The first would tend to use brighter lights as
anchors.
> What's wrong with this picture and my calculations?
>
> Chuck Norcutt
>
> AG Schnozz wrote:
>
> > Chuck Norcutt wrote:
> >
> >>Yes, digital image stabilization which is derived from
> >>camcorder technology and probably requires a real time
> >>image. I don't think that's likely for an E-x machine.
> >
> > I'd love to be vindicated on this one. I wrote months and months
> > ago that I expected Olympus to go image-stabilization digitally
> > instead of mechanically.
> >
> > Yes, I'd buy it. Just give me 10+ MP.
> >
> > Nice thing about Olympus with the E-thingy pro line. You can
> > buy the next model and not worry about it getting obsoleted by
> > Olympus for at least three or four years.
No, other makers will obsolete it...
tOM
-- Quidquid latine dictum sit altum viditur --
,__@ tOM Trottier
_-\_<, 758 Albert St., Ottawa ON Canada K1R 7V8
(*)/'(*) N45.41235 W75.71345 +1 613 860-6633
<a href="http://Abacurial.com";>Abacurial Information Architecture</a>
Q, Q,
</ </ This world, after all our science and sciences, is still
(`-/---/-') a miracle; wonderful, inscrutable, magical and more,
~~@~~~~@~~~~~~ to whosoever will think of it. --Thomas Carlyle
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|