Engineers from ESPN were on hand at the GPU Technology Conference to share how the network is using GPU-powered software architecture to enhance the viewing experience.
“Just about every event that goes on the air, we’re trying to get new things in to enhance the viewing experience,” says principal engineer Mark Muench.
During ESPN’s coverage of the Home Run Derby, for example – held in conjunction with Major League Baseball’s All-Star Game – GPUs helped the broadcaster pop up a real-time graphic that shows where each home run lands and how it’s traveled.
To accomplish this, ESPN captured four separate HD channels that fed into its system and are then layered over a model of the stadium on a single machine.
The network also has relied on its GPU architecture to support an ESPN3 broadcast of simultaneous youth football teams, showing the separate feeds in a multi-screen display that helps the crews react in real-time, putting the most pressing action on the biggest of the three screen areas.
Additionally, ESPN uses its GPU architecture to create a workflow called Snapzoom that converts 4K ultra-HD inputs into a more standard 720P signal, and enables lossless zooming into the image, still in HD.
What’s more, software engineer Chris Bond said the network is working on a similar workflow that will let it create a live virtual zoom camera and then pan all over the displayed image while zoomed.
And things figure to get even more mind-blowing soon. ESPN is upgrading its underlying NVIDIA architecture from Fermi to Kepler, and is already seeing a 100 percent improvement in performance as a result of being able to support 4K signals without requiring a second GPU, as is now the case.
That doesn’t mean ultra-HD sports programming will be coming into the home any time soon. But Muench said 4K is figuring more prominently in ESPN’s plans, thanks in large part to its GPU architecture.
“Right now, we’re looking at how we can use 4K to enhance the delivery of ESPN for the home viewer,” he said.
Of course, the next generation of NVIDIA technologies could very well accelerate that strategy.