Visualising the future
Visual computing - a long way to go
At a session on the advancements in visual computing at Nvidia's GTC, academics galore flexed their mental muscles, pontificating on the best way to rid themselves of pesky problems arising from segmentation, correspondence, recognition and more. Riveting stuff indeed!
"There is a considerable amount of improvement we still need to achieve" explained Austrian professor, Horst Bischoff, from the Institute for Computer Graphics and Vision at the Technical University of Graz, adding that a plethora of problems still needed sorting before visual computing became an integral part of the big picture.
The pervasive feeling at GTC, however, is that, despite the challenges, answers are on their way, especially thanks to recent advances in hardware.
Rome built in a day
Bischoff pointed to a recent project which aggregated 2,106 photos of the Colosseum in Rome, posted on Flickr, consisting of some 819,242 points, all melded together into a rather stunning 3D visual model which took just 21 hours to compile. Cue the clichés about building Rome in a day.
Aside from adding to PC power, the tech-sperts waxed lyrical about the potential for augmented reality on mobile devices and handsets, especially by streaming through the cloud, allowing visual computing to become a lot more accessible to the unwashed masses rather than just scientists and supercomputer aficionados.
Not about graphics
Ironically, Canon professor of computer science and electrical engineering at Stanford, Pat Hanrahan, said visual computing wasn't even so much about the actual graphics, but more about better visually representing massive databases in n-dimensional images. Hanrahan, a former Pixar employee, delivered the astounding statistic that every person managed to produce some 2GB of information a year, and retail giants like Walmart had to surmount the challenges involved in storing 1/2 petabyte (500 terabytes) of data per year, making the need for some kind of better ‘visual' computing incredibly pressing.
Hanrahan said that although he didn't yet have all the answers, he was incredibly excited by the prospect that sometime soon visualization would becoming available anywhere and everywhere simultaneously, and that manipulating and interacting with data would be as simple as playing with Google Maps.