Originally posted by dms Seems to me there are too many arbitrary assumptions. For one why a 1 pixel blur? if this is the criteria we are not doing photographs where SR is a consideration. Ditto a 1 second exposure. And even w/o a detailed analysis intuition suggests this explanation/science is (very) implausible.
The point is that if you read the data from the gyroscopes inside a camera sitting on solid ground, that data will say that the camera is slowly spinning (because the Earth is turning). The SR system will try to remove the effects of that measured spin on the image. If one is taking a picture of the stars, removing the spin of the Earth is great (that's what Astrotracer does). But if you are taking pictures of landscapes, buildings, people, etc. who are also spinning with the camera once every 24 hours, the SR is adding blur to the image.
The amount of blur (in pixels) added by the SR system's "misinterpretation" of the gyroscope data depends on a lot of factors (sensor pixel size, focal length, exposure time, and pointing direction). For a 20 MPix M43 camera (e.g., the latest Olympus camera), the added blur would reach 1 pixel (worst case) for a 100 mm lens after 0.6 seconds or a 200 mm lens after 1/3 of a seconds.
Now whether 1 pixel of added blur is noticeable or unacceptable really depends on the subject matter and the photographer. But that subjectivity is always going to be an issue with both the 1/focal length rule of thumb for handheld photography as well as different photographers experience with how many stops SR actually provides.
The deeper point is that the motion of the Earth does affect SR by measurable amounts for telephoto lenses although the subjective effects are arguable.