Euclid Discoveries is shopping a software modification that it has demonstrated makes MPEG-2 encoding anywhere from 10 percent to 30 percent more efficient.
The first step of MPEG-2 (and MPEG-4) encoding is predictive modeling – anticipating which macroblocks will change from frame to frame, and how. Euclid has refined predictive techniques to get beyond the macroblock level, instead employing a means to identify and model objects in the video frame.
The company says the benefits of the technique are actually more pronounced when encoding scenes with rapid motion.
Like all efficiencies in compression, Euclid’s technology provides video distributors the option to either increase compression rates to save bandwidth or to increase the quality of the video.
Euclid Discoveries’ co-founder and CEO Richard Wingard said the company has no interest in becoming an encoder supplier. Instead, Euclid is looking to license its technology to encoder companies.
Euclid has a software module currently available for MPEG-2 encoders, and expects to have one for MPEG-4/H.264 by the end of the year. It is working to have software developer kits for MPEG-2 encoders ready soon, and SDKs for the MPEG-4 version when that version is released. The technique is also applicable to HEVC compression.
To get encoder companies interested, Euclid has gone straight to the MSOs, hoping MSOs will get enthusiastic about it and begin requiring the technology.
MSOs have MPEG-2 set-top boxes installed in tens of millions of viewers’ homes, set-tops that operators expect to last for another 8- to 10 years. And so MSOs understandably might be interested in proposals that would support their investment.
According to Wingard, that strategy is beginning to work. He claims his company is getting serious interest from two of the top three cable operators in the U.S.
If the average MPEG video stream could be compressed further, say from 11 Mbps to 8 Mbps, for example, operators would be able to get four streams through a multiplexer where once they could fit only a maximum of three, Wingard said.
Wingard said that encoders equipped with his company’s predictive modeling module have been measured by BD rate and visual comparisons, which is how the company arrived at its estimation of compression gains of between 10- and 30 percent for high motion, high definition videos.
Wingard promised that future versions of the technology for MPEG-2 encoders will produce additional compression gains and will be targeting enablement of real-time video processing. These additional techniques are appropriate for low motion, high ambiguity video – scenes with higher levels of visual complexity.
Euclid Discoveries is shopping a software modification that it has demonstrated makes MPEG-2 encoding anywhere from 10 percent to 30 percent more efficient. Euclid has refined predictive techniques to get beyond the macroblock level, instead employing a means to identify and model objects in the video frame.