Do you have in-camera or lens-based anti-shake?
How much do you rely on it?
Does it really give you 3 stops more, as advertised?
I just checked my photos and found some very sharp ones, shot at 1/20th with Pentax' in-camera shake reduction. But others, at 1/45th, were not sharp.
So I really wonder if this really works that well. Or if one gets sloppy, because "there's anti-shake".
For my new camera set-up (planned: Canon 40D) I have the option to get a lens without IS (lens-based anti-shake) or, much more expensive, the 17-55/2.8 IS (which I'd rather not buy because of its inclination to suck in dust, and it's 1000$ for non-L glass). The competing lens would be the Sigma 17-70. Unfortunately, there's not much choice for the cropped sensor. No 24-105 L for APS-H sized sensor.
[{POST_SNAPBACK}][/a]
View,
Your post confuses me. The Canon 40D has an APS-C sensor, not APS-H. Not sure what you mean by "not much choice for the cropped sensor." All EF lenses work with the APS-C sensor. However, EF-S lenses are specially designed for APS-C based cameras and will not work with full-frame sensor-based cameras.
Your experience with anti-shake technology correctly reflects the statistical nature of the technology. In the case of the Canon in-lens IS system, the gyros sense your handshake and command one or more lens elements to move to counteract your handshake. However, the movable lens element cannot move in the same direction forever. After it hits its stop (the physical limit of its angular adjustment), or between shots, the adjustable lens element must be re-centered by the IS system. If you take an exposure at this time, while the IS system is re-centering and is thus unavailable, you may get a blurry image depending upon shutter speed, etc. Statistically this "down time" will be more likely to coincide with an exposure made with three stops slower shutter speed than with one stop, for example. Thus, the more you push the IS to save you the more likely it will be that you get a fuzzy image.
None of this means that IS is not worth every penny, depending upon how much you value your photography, because it will save MANY shots. I would never even think of buying an EF lens in a non-IS version if an IS version is available. The only situations that I can think of to justify doing so are (1) an extremely tight budget, (2) photography where no camera shake is guaranteed (e.g., tripod-based), or (3) guaranteed light sufficient to expose properly with high shutter speeds (e.g., in a portrait studio with strobes).
I would also recommend that you buy an L lens if you can possibly afford it.
For more information on the statistical nature of the IS system, see:
[a href=\"http://www.dpreview.com/lensreviews/canon_70-200_2p8_is_usm_c16/page5.asp]http://www.dpreview.com/lensreviews/canon_...m_c16/page5.asp[/url]
That URL is to a page from a DPR review of the EF 70-200 f/2.8L IS. Page down to the section "Optical Image Stabilization" to see a statistical bar graph showing the percentage of time that the DPR tests yielded a particular amount of sharpness at various shutter speeds.
Good luck,
Bruce