Hi, very interesting indeed but...
Mike Liebhold schrieb: > 4. We better move quick, The Handheld AR meme is in the air. The > Tonchidot demo demonstrates, that the mainstream techcruch VC crowd is > now dialed and very stoked by the possibilties the viewfinder as the > next browser. > > > to me this Tonchidot demo shows at first, that it really isn't working now. I didn't got (and I think the presenter of this demo too), if they get the information for the tags from markers on the objects which should be tagged, or if they "did" geometry matching with 3D models or 2D pictures of the objects? The marker thing should be relatively easy to implement on actual hardware, but the difficult (at some point impossible) thing of this approach would be the tagging of all the objects. :) The geometry matching approach needs much cpu power (in my oppinion to much for actual mobile devices to achieve a good response time) and also huge databases of 3d modeles of the to be tagged objects. For tagging geolocated objects you could apply spatial filtering to search smaller amounts of models/geometries to match, but I think this would still need a big amount of processing and also big databases. Or is there a faster/better/other technology out there to do what Tonchidot claims to be able to do? In the moment I just can't see a way to do that application on mobile hardware of the next approx. 5-10 years!? curiously Christian _______________________________________________ Geowanking mailing list [email protected] http://lists.burri.to/mailman/listinfo/geowanking
