Hi, I brought up this shift function with APS SENSORS/FF lenses a while back, I don't see why it wouldn't work but you have not addressed the addition burden of needing a FF viewfinder for it to be workable and some sort viewfinder indication to show what portion of the viewfinder image the sensor is capturing...
As for whether it's a viable product or not its all about cost. If it could be done for only slightly more that plain ol APS it might be viable. The other thing is that your good idea about covering the whole frame is only possible with stationary subjects while the shift function doesn't have that limitation. The stationary subject requirement makes it very little better than Pan and stitch techniques already possible with any digital camera... JCO -----Original Message----- From: Glen [mailto:[EMAIL PROTECTED] Sent: Thursday, September 22, 2005 3:18 AM To: [email protected] Subject: Sensors That Shift? Since the current sensors are smaller than 24 x 36 mm, could a camera be built in such a way that it could shift the sensor up and down, and from side to side? This would have a similar use as a shift lens would have. Of course, you would have to use lenses made to cover the full 24 x 36 format. The current DA lenses wouldn't work with a moveable sensor. I think this might be a cool feature for some people, especially for those who photograph architecture. There is also one other potential use for a moveable sensor. When photographing stationary objects and using a tripod, the sensor could be used to take more than one image of the subject. Each image would be taken with the sensor shifted a sub-pixel distance vertically and horizontally between images. Let's say that instead of a single image, we capture 9 images, arranged in a grid pattern centered around what would have been the normal single image. We then use this grid of 9 images to create a higher-resolution image than a single image capture would have produced. This should be a way to quadruple the amount of effective pixels, while using the same sensor. Unfortunately, it would only work for situations where the camera and subject were kept stationary with respect to each other. Still, I bet a lot of photographers would benefit from such a boost in resolution under such circumstances. Also, you could possibly develop some enhanced noise reduction techniques by analyzing the extra exposures and tossing out any pixels which seemed abnormally bright. Would this idea of shifting the sensor in sub-pixel amounts actually help yield higher resolutions? My intuition tells me that it would give images with higher effective resolution than the single images we currently have, but perhaps not quite as nice as a sensor with a truly quadrupled pixel count. However, it should be a lot cheaper to build than a sensor with a quadrupled pixel count. ;) Of course, the sub-pixel shift was thought up as a way to boost effective resolution, while maintaining full compatibility with DA lenses. If the sub-pixel shift idea doesn't work, I have a second idea for increasing the total resolution of the captured image. Once again, it only works for stationary subjects. For each image, make 4 captures. Shift the position of the sensor to the top-left corner of the 24x36mm frame for the 1st capture, then to the top-right corner, then the bottom-right, and finally to the bottom-left corner of the 24x36mm frame. This way, you have covered the full 24x36mm frame in 4 tiles. Then, seamlessly stitch these tiles together in software to create a single full-frame image with much more than our normal 6.1 megapixels. Of course you lose the compatibility with DA lenses, and would have to use lenses designed to cover the full 24x36mm frame. I'm not sure how many extra megapixels you would gain from simply extending the effective sensor coverage to 24x36 mm. Maybe someone on the list knows how to calculate this? take care, Glen
