[johnmflores]

It is a marketing/sales feature and therefore will never go away.

Photographers have always looked for an extra stop. And they always will.

[Lowell Goudge]

In the case of the Pentax SR system (the one I studied) it is gyro sensors, not accelerometers. The gyro sensors (there are two in the K20D) work by measuring the corilios force and are pretty ingenious pieces of electonics
In some of my earlier posts, I gave the exact manufacturer and part number.

interesting, I would have assumed accelerometers since they would also allow for the implementation of the level and auto rotation of images when you tip the camera on the K7. otherwise there is an additional 3D accelerometer for this because level is static and a gyro would not be able to measure it.

Wrong. This is the nice part about sensor-based stabilization. You only need to know the angular frequency around the optical axis and the two axis perpendicular to it and the focal length for an exact compensation of an image at infinite distance. Lens details play no rôle here.

At finite distances, you need extra accelerometers and Canon did actually file a patent explaining the obvious.

I would have to disagree not perhaps with your statements but the issue at hand.

I don't know how others shoot, but my shots are very very rairly at infinity. as a result, while you may be correct in your statements, for other than infinity you need to determine if shake is rotational around some point, or purely lateral,

That is what I was getting at, and since the center of mass and the theoretical location of the objective (i forget the term) in each of my 3 examples is quite different, in the real world the SR needed is lens dependant. I don't use my 500mm to look at a flag 2 miles away, I use it to look at a 10-15 cm high bird 10 meters or less away and fill the frame with it. To that end, however, as my first post indicates i like the performance

[falconeye]

I don't know how others shoot, but my shots are very very rairly at infinity. as a result, while you may be correct in your statements, for other than infinity you need to determine if shake is rotational around some point, or purely lateral,

That is what I was getting at, and since the center of mass and the theoretical location of the objective (i forget the term) in each of my 3 examples is quite different, in the real world the SR needed is lens dependant.

Well, you either measure the translational degrees of freedom (i.e., additional accelerometers) or not. If you do, you compensate for them. Which again factors out lens details (like center of gravity). I do concede however that you would benefit from knowing the nodal point distance from the focal plane which is only approximately equal to the focal length.

Or, you don't measure the translational degrees of freedom and only measure rotational degrees of freedom. Which again takes any opportunity to make use of lens detail knowledge.

AFAIK, all vendors to date base their IS on rotational degrees of freedom only. Which is an excellent simplification if subject distance is much larger than focal length which, except in Macro photography, always is the case.

In your 500mm 10m bird example, a rotation by 21 arcsec only is as bad as a lateral shift by 1 mm (namely 10 pixel blur). And 21 arcsec is little (1/175°)! Much more difficult to keep than to keep a position to a millimeter.

[Lowell Goudge]

Well, you either measure the translational degrees of freedom (i.e., additional accelerometers) or not. If you do, you compensate for them. Which again factors out lens details (like center of gravity). I do concede however that you would benefit from knowing the nodal point distance from the focal plane which is only approximately equal to the focal length.

Or, you don't measure the translational degrees of freedom and only measure rotational degrees of freedom. Which again takes any opportunity to make use of lens detail knowledge.

AFAIK, all vendors to date base their IS on rotational degrees of freedom only. Which is an excellent simplification if subject distance is much larger than focal length which, except in Macro photography, always is the case.

If you consider the nodal point is at the front element of the lens and independant of lens design, while that may be true for a simple lens such as a 35-100mm lens, and it may also be very close to true for an SMC 500F4.5, it is clearly not the case for a true telephoto (telepohoto by definition defines a lens that is physically much shorter than the equivelent simple optical design, and achieved by ading additional elements) or for a mirror where the nodal point may be at the front element but it is only 100-150 mm away from the camera body not 400-500mm.

As I said, i believe this is a point where I think all I can say is I agree we disagree.

[falconeye]

If you consider the nodal point...

I already conceeded that additional knowledge about the nodal point would be beneficial. Provided that there are additional accelerometers as well.

[Clinton]

I don't think there comes a point at which we can just do away with SR or IS.

Even if the camera can suddenly do so much iso pristinely that it can shoot F32 in pitch black at 1/8000th and make a fantastic image, there will be cases where we may want a slower shutter speed to create a specific look. In those cases we'll still want SR, and we may want better SR than we have now.

[WMBP]

I don't think there comes a point at which we can just do away with SR or IS.

Agreed. On the other hand, what does seem possible, is that, at some point image stabilization becomes so universal that it simply isn't any longer a marketing feature of great value.

We may just about be there, now that Nikon at least is sticking VR lenses even on its low-end consumer DSLRs. I suspect at this point that other virtues—such as auto-focus performance and accuracy or high-ISO, low-light performance—are, for many buyers, more important selling points than image stabilization.

Will

[falconeye]

The more I think about it, the more I believe that today's IS is the tip of an iceberg only.

My next camera (or the one after...) will feature the ultra-freeze IS technology: the camera will freeze in space and take ultra-steady shots. And make tripods obsolete.

It don't care how. Let's just assume it hooks itself in subspace

[lurchlarson]

SR or IS is still very important. Even though ISOs can go really high there are still times when you need the smoothest image possible and high ISO can always deliver. Therefore, SR is still needed.

[leok]

Image stabilization will always be handy, it allows you that steady shot without changing shutter speed, its allows shots to be taken in circumstances where you would normally need a tripod.

High ISO is also handy, but using it means you have a different set of shutter speed & aperture than you would get with a lower ISO and IS.

Therefore IS and ISO compliment each other, and allow shooting with a wider range of settings than either would alone.