Re: Nikon maintains DSLR lead over Canon

David J Taylor wrote:
Alan Browne wrote:
David J Taylor wrote:
Alan Browne wrote:
[]
As I stated a very long time ago, the main failing of in-camera
shake reduction is that it cannot compensate for pitch/yaw movement
which is the greater contributor to image blur. In camera helps,
but cannot be as good as optical systems.

Cheers,
Alan
Could you expand on that, Alan, please? I don't currently
understand your reasoning. I can see that lens-based IS could offer
a greater shift than a sensor-shift system.
The lens based IS/VR [Canon/Nikon] move an optical element (double
concave IIRC) in the x,y axis' (the lens axiz is z) to effect PITCH
and ROLL corrections in the movement. These are the two greatest
contributors to image blur from photog movement.

The sensor "moving" systems always have the sensor perpendicular to
the z axis and are attempting to keep the image in one spot while the
camera moves in translation in the x and y directions. The system is
correcting for a lesser contributor to image blur.

What Pentax have added is a rotation around the lens axis (not tilts
as Tony suggested) and this is a useful improvement as another
contribution to blur is camera rotation around the lens axis due to
less than gentle shutter depress. The only drawback (it just occured
to me) is that the system would have to measure this and compute the
amount of correction as it happens whereas the x,y movements are
measured before the shutter fires.

Cheers,
Alan

Thanks, Alan.

I don't quite follow your analysis for the sensor-moving systems. You seem to be suggesting that accelerometers are measuring the camera's X and Y translation, rather than the rotation which I'm sure they are actually sensing. Any translation in the camera would only affect quite close subjects, at least compared with the effects of rotation.

The detections, per the Pentax data sheet sense rotation (I take this to

http://www.pentaximaging.com/files/scms_docs/SHAKE_REDUCTION_FACT_SHEET.pdf

mean pitch and yaw and probably roll) but they are actuating the image sensor in translation (x,y) and lens axis rotation only. Making assumptions (perhaps based on the mounted lens) on center of gravity could provide a reference point for computing the translation vectors from the rotation vectors. (This last bit is speculation, the data sheet mentions that the system will perform best on lenses for which the camera can read the focal length or for which the user enters the focal length (older lenses)).

These systems don't detect movement based on the image content, so the brief absences of the image wouldn't matter for either the rotations along the X- and Y-axis, nor for rotations about the Z-axis.

I wasn't implying that it was sensing the image (hard to do with the mirror down and the shutter closed) but rather that based on the inertially sensed movement it is attempting to keep the moving sensor centered on the desired image location. (No different than the lens systems which based on movement are trying to project the desired image to the right location on the still sensor).

Like the K-M system, Pentax do not recommend the S-R for panning shots. This makes wonder if they could isolate the Shake function in vertical only and freeze the horizontal. This could possibly even be sensed and automatically done. (As it is in some second gen IS lenses; not sure about VR).

I've not been aware of actually incurring rotational camera shake, but perhaps I'll find it if I look more carefully.

If your shutter release technique is good, then it is not much (any) of an issue. But many people have poor "jab" technique. As the shutter is way off lens axis, a rotation and backlash will occur. I still find myself occasionally lapsing into poor shutter release technique.

Mount a laser parallel to the lens axis to see.

Cheers,
Alan.

--
-- r.p.e.35mm user resource: http://www.aliasimages.com/rpe35mmur.htm
-- r.p.d.slr-systems: http://www.aliasimages.com/rpdslrsysur.htm
-- [SI] gallery & rulz: http://www.pbase.com/shootin
-- e-meil: Remove FreeLunch.
.

Relevant Pages

  • Re: Nikon maintains DSLR lead over Canon
    ... system of any DSLR. ... improvement, it also tilts the sensor slightly, better aligning it ... _rotates_ around the lens axis. ... directions of camera movement contribute what amount to camera shake visible ...
    (rec.photo.digital.slr-systems)
  • Re: globe trotting?
    ... // First we do the look up and down rotation on the x axis. ... consider storing only a camera origin ... > where R is "big enough" to put the camera outside the sphere. ...
    (comp.graphics.algorithms)
  • Re: Viewpoint rotating around a sphere
    ... which lets you actually position the camera (in ... If I take rotation out of the equation altogether, it works fine, ... Basically I'm drawing a bunch of atoms and a bunch of bond ... That all works fine if I use the Identity matrix as the starting point for ...
    (comp.graphics.api.opengl)
  • Re: globe trotting?
    ... A good camera scheme has a minimal set of "just right" independent ... Assume the world origin at the center of the sphere. ... of the whole camera about the world origin on an axis perpendicular to ... The angle of the rotation is a function of walking speed, ...
    (comp.graphics.algorithms)
  • Re: globe trotting?
    ... A good camera scheme has a minimal set of "just right" independent ... Assume the world origin at the center of the sphere. ... of the whole camera about the world origin on an axis perpendicular to ... The angle of the rotation is a function of walking speed, ...
    (comp.graphics.algorithms)