It only works to a point Alan. If something is too far out of focus, the details are quite simply GONE, you may return the shape of some objects, but surface details that would make it a printable photograph will be completely and utterly lost... You can return some larger details in slightly out of focus imagery, but even then, if too much data is lost due to poor focus, IT CANNOT BE MAGICALLY RECONSTITUTED... Even if you know exactly how a lens blurs an image, and the subject distance, you cannot take a picture that is out of focus beyond a certain point and get any details back.... You just can't. I am familiar with deconvolution, I've heard of it being used to enhance video on security tapes... In slightly blurred images, deconvolution works very well, it is much better than unsharpen mask for recovering slightly blurred photos. But like I said, you aren't going to magically reconstitute lost data no matter what. If you can take a severely blurred photo and reconstitute the shapes and details I will eat my words, but until then I stand by my original statement. When things are out of focus, light rays overlap, the first things to overlap are details, once the details overlap, they're essentially gone. In a badly blurred image at any pixel site you've got light from neighboring details intersecting with light from other neighboring details and the end result is a color totally different from what should be there, and only god knows what combination of scattered light rays created that color.
-That Guy -----Original Message----- From: Herb Chong [mailto:[EMAIL PROTECTED] Sent: Tuesday, June 15, 2004 7:36 PM To: [EMAIL PROTECTED] Subject: Re: IS in *istD shows how much you know about deconvolution. Herb.... ----- Original Message ----- From: "That Guy" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Tuesday, June 15, 2004 1:12 PM Subject: RE: IS in *istD > If an image is out of focus, data is lost, and cannot be recovered by any > sort of math... If in an image is sharp, yet distorted mildly by > pincushion or the like, there is enough data in the image to correct the > image and suffer little or no quality loss.
