4 ms·
I wonder if, rather than storing an "image" of a face, we store a sort of error correction from a standard model. This is akin to probabilistic compression, wh
by daeken 13y ago
I wonder if, rather than storing an "image" of a face, we store a sort of error correction from a standard model. This is akin to probabilistic compression, where you have a bitstream generator and then store a list of where that bitstream is in error.
- zk00006 13y agoNobody knows how brain works exactly, but I believe that some sort of compression is there. This has been done in computer vision some time ago and there is a nice application which essentially represents a face by 100 dimensional vector: www.facegen.com. Starting form average face, its possible to get any face you like by varying those dimensions.