would it be possible to do it by using one of these two approaches after the first general iPad user agent check?

-using a full screen image and getting an useful parameter from that via javascript

-uploading a screenshot of a graphics element via javascript/canvas for a server to evaluate