On 4 Jan 2006 at 12:16, Tim Øsleby wrote: > Knowing the FL makes it lot easier to make a IS system. But I do believe it > is possible to make a more "intelligent" solution. > My idea is something like this: > First it measures the camera movements, and makes a rough guess. Then a > feedback system (based on data from the image sensor) tells the camera how > successful the first estimate was. Based on that information you will have > data for a better calculation. (A similar technique has been used with some > success in subwoofers in sound systems).
Yes I've seen sub-woofer systems with accelerometers on their cones which use a a near real-time closed feedback system to reduce distortion but unfortunately we don't have such a direct mode of feedback when compensating for camera shake. And even if we did the AS system would have to make assumptions regarding subject relevance as does an AF system. > This is just my non expert speculations... > What I'm trying to say is: We don't _know_ that a IS system needs the FL, > you are just assuming it. We do know it needs to know FL because the image being shot isn't being monitored real time during the exposure. The only way that a measured movement of the body can be translated to offsets at the focal plane is if the lenses FL is known. Rob Studdert HURSTVILLE AUSTRALIA Tel +61-2-9554-4110 UTC(GMT) +10 Hours [EMAIL PROTECTED] http://members.ozemail.com.au/~distudio/publications/ Pentax user since 1986, PDMLer since 1998
