Alan Chan wrote: > To achieve IS/VR/AS, the camera or the lens must be able to detect their own > movement, and obviousely *istD doesn't have any. For this reason, the > firmware idea doesn't work.
Alan, I don't know why the camera or lens would have to detect movement. That would require some type of accelerometers and would be a finicky cludge IMO. Software implementation of image stabilization might be achievable by: 1: taking a high speed sensor reading defining only high contrast areas (in effect a super high ISO reading of only light dark edge lines) 2 taking another longer duration sensor reading to determine some secondary lower contrast areas 3. taking the lowest sensor reading for color. 4. Combine the three readings in camera and you would have an image stabilized picture. Another method would be to capture a series of quick high ISO readings, calculate movement through interpolation of the images captured, take a lower speed reading for color and texture combine them all through interpolation and save the result to memory. Textures might present a problem with that method. Software image stabilization would be a compromise but it's conceivable (in my mind anyway). Tom Reese
