Nvidia Quadros go for Olympic gold

Posted

With the winter Olympic fever at an all-time high, Nvidia is doing its best to cash in on some Vancouver gold, plugging its Quadro cards as having enabled NBC to display all sorts of neat feats and computational wonders to enhance your favorite snow sports.

Teaming up with Sportvision, the firm which does the “RACEf/x,” real-time info ‘balloons’ that track the cars at NASCAR events and the superimposed first-down marker at NFL games, Nvidia says its pro GPUs are also helping with a new effect called “SimulCam.”

SimulCam apparently superimposes one athlete’s performance over another to graphically illustrate the differences between competitor’s strategies, approaches and even their flaws.

For instance, says Nvidia, SimulCam allows viewers to compare one skier’s performance against another’s simultaneously. 

Another technique called “StroMotion,” is also being used, repeatedly freezing [no pun intended] athletes in motion during a given segment of their routine to demonstrate - within a single frame - the entire evolution of their movements.

Both visual technologies were actually developed by a Swiss firm, DartFish, which had originally intended to use them for athlete training.

The tech behind the Olympic visuals is actually, again no pun intended, quite cool.

Both work by compounding video images into a frame-by-frame sequence, but as the name implies, StroMotion is based on stroboscoping – ensuring a moving object is perceived as a series of static images along the object’s trajectory – whilst SimulCam is a video processing application combining video sequences with Spatial-Temporal alignment.

“Given two video sequences, a composite video sequence can be generated which includes visual elements from each of the given sequences, suitably synchronized and represented in a chosen focal plane,” says Nvidia’s press release on the subject. 

SimulCam also uses a fair bit of background recognition, which identifies the pixels that belong in the background and calculates how those pixels move throughout a series of successive images. Rather compute intensive, and something Nvidia reckons is made a lot easier using its Quadros.

Meanwhile, StroMotion computes the camera movement between every two successive video images. Once determined, it stitches the images together, and is actually able to remove the moving object from the image. 

Such is the awesomeness of GPU acceleration that StroMotion can also determine how each video image relates geometrically to each other and to the panorama.

All a bit complicated for us. We’re off to numb our brains a bit by watching the hockey.