NAB Perspectives: NVIDIA’s Justin Boitano Sees GPU Technology as Pathway to High-Octane Graphics Workflows
When dealing with on-air graphics and image processing, broadcasters are often forced to use lower-quality images in order to get the product to air with minimal latency. NVIDIA aims to change that by introducing its graphics-processing–unit (GPU) technology to a wide range of I/O devices at the NAB Show.
“All the video-I/O-card companies in the industry want to get low-latency access to GPUs because they are so adept at performing video processing,” says NVIDIA Director of Marketing Justin Boitano. “By combining their great I/O with our GPUs, you get an amazing platform for future innovation and faster workflow. There are countless examples of image-processing work that is phenomenally well-suited for GPUs, and we’re just looking to provide that.”
NVIDIA has already partnered with several major I/O-board manufacturers — Blackmagic Design, Bluefish444, Deltacast, DVS, Matrox — on its GPUDirect for Video solution, which uses NVIDIA Quadro and Tesla GPUs to provide ultra-low-latency input and output for I/O devices. NVIDIA provided these I/O-board vendors with an SDK at IBC 2011 and expects to see application developers actually bring products to market by next year’s NAB Show.
“By this time next year, this is going to be a huge trend,” says Boitano. “The I/O-card companies will soon provide that SDK to application companies.
NVIDIA GPUDirect for Video technology is designed to get video data quickly in and out of the GPU. Software vendors are now capable of harnessing the graphics- and image-processing power of GPUs without the latency previously associated with third-party video I/O boards.
The technology eliminates delays of as many as 10 frames — an amount easily visible to the naked eye — when transferring video from an I/O device to a GPU, by enabling synchronized communications between the two. Developers no longer have to manage complex buffering schemes, which historically have led to unnecessary CPU overhead and increased latency.
Boitano notes, “A Vizrt type [of company] can take advantage of this unified platform and go to CNN for election coverage or a sports network for live event coverage and offer a very low-latency way to get video into the GPUs, which allows [CNN] to use virtual sets and virtual elements in new ways that we could never have even dreamed of before because it just took too long for live television. Once it gets through that adoption path, this will change a lot of [the current broadcast workflows] for graphics.”
NVIDIA’s GPU technology may be a year away from entering broadcast workflows, but Adobe Creative Suite users have already experienced its benefits, because Version 5 of the production suite was the first to incorporate NVIDIA Quadro and Tesla professional GPUs. This continues with the upcoming release of Adobe Creative Suite 6 Production Premium. According to NVIDIA, motion graphics that would take hours to create can now be done in minutes, and effects that used to render a few frames per second now run at full speed, and high-quality color grading is now possible, thanks to NVIDIA GPUs. CLICK HERE for details on NVIDIA’s role in CS6.