There's an idea that I had that I just wanted to get out in public so that no one can patent it:
to encode a polygonal mesh as the union of a) separately encoded connectivity w/o geometry; and b) an unordered cloud of points to provide geometry w/o connectivity. The two would then be married with prediction algorithms typical of connectivity compression, with small corrective factors along the way.
My plan was to resolve vertex id's in the same order that the standard edgebreaker traversal discovers them. Any new vertex that didn't have a full parallelogram to do its completion with
(which would, of course, include the very first triangle), would be encoded with an explicit id in the point cloud. All others would predict as before, and then the encoding would be something
like "take the 5th closest vertex to your assumed location".
Slashdot Mirror
There's an idea that I had that I just wanted to get out in public so that no one can patent it: to encode a polygonal mesh as the union of a) separately encoded connectivity w/o geometry; and b) an unordered cloud of points to provide geometry w/o connectivity. The two would then be married with prediction algorithms typical of connectivity compression, with small corrective factors along the way. My plan was to resolve vertex id's in the same order that the standard edgebreaker traversal discovers them. Any new vertex that didn't have a full parallelogram to do its completion with (which would, of course, include the very first triangle), would be encoded with an explicit id in the point cloud. All others would predict as before, and then the encoding would be something like "take the 5th closest vertex to your assumed location".
He was probably referring to fortune cookies. see here for a history of fortune cookies. http://www.chinatown-online.co.uk/pages/food/fortu ne.html
Couldn't this be used to help contain sars? I mean, keeping track of who came in close contact with whom is the hardest part