How far does SSS have to move the sensor (warning - very long)

Part 1 - the problem (split to avoid exceeding message size limit)

The debate over the relative merits of sensor versus lens-based image stabilisation is surely going to be one of those perpetual photographic controversies such as CCD vs CMOS, film vs digital and manual vs automatic. Add your own particular examples to the list. Some differences are inarguable – systems like SSS can’t stabilise an optical viewfinder image. Other issues are more open to debate. Having a vested interest in a bundle of A-Mount kit, I decided to conduct a little test to see if there is much truth in the stories that sensor-based stabilisation is inherently incapable of coping with very long focal lengths or full-frame sensors. This little experiment doesn’t directly answer these points, but it does indicate where the limits might be.

As everybody knows, SSS works by moving the sensor in the X & Y planes (down/up, left/right) to compensate for angular deflections of the camera (and arguably lateral ones too, but these are generally inconsequential). It doesn’t compensate for front/back movements, but that’s only significant for circumstances when there’s a very shallow depth of field (typically macro-photography where you’d probably use a tripod anyway).

The charges directed towards sensor-based stabilization are essentially related to the degree of movement that has to be achieved in a given period of time. Firstly the magnitude of the movement is a product of the angular deflection of the camera and the focal length of the lens. Broadly, the longer the focal length, the further the sensor has to be moved in response. Taken to extremes, the sensor would be moved outside of the imaging circle of the lens and this could prevent the use of true full frame. Further, this movement has to be achieved very quickly and precisely, which is a major engineering challenge and potentially power-sapping.

So this little experiment, which is easy to repeat, is just an attempt to assess the amount of lateral movement the sensor has to accommodate to maintain a sharp image. I’m not claiming this is a fully controlled experiment, but I think it good enough to get some reasonable estimates. The test was performed using a 500mm lens on a Sony Alpha A700 but could be repeated on any reasonably long lens.

see part 2 - test and results