This deals with the (diverse) flash viewers that let you zoom in on images on websites. I’m trying to extract the large, zoomed-in image rendered by the viewer. In many cases the images seem to be dynamically called by the viewer, or are created only for the part of the image you are zooming on at that point. Ideally, the approach here would be a programmatic one that could be called on an identified flash element. Expect there is nothing universal, but interested in the top few approaches that will grab most cases.
Use an http monitor (firebug has one). This will give you the URLs of the image tiles. Then, it's down to you to figure out the pattern in the URLs that will give you the tiles you need.