In an indication of the evolution of the video infrastructure industry, V-Nova has added machine learning capabilities at the core of its flagship Perseus Pro codec, which powers the P.Link Dual UHD/8HD video encoder and decoder.
The video compression solutions provider says the addition represents the deepest current application of machine learning in its field. Available for integration into contribution or production workflow, the P.Link encoder/decoder hardware is designed to enable the TV and media industry to realise cost savings in bandwidth or quality upgrades across a range of use-cases including remote production, live event contribution and site-to-site links. P.Link also allows high density HD and UHD bi-directional flexibility, featuring dynamic multiplexing to further optimise the delivery of multiple concurrent feeds, and includes HDR support as standard.
By incorporating machine learning and convolutional neural networks into its multi-scale approach, Perseus Pro has been engineered to identify common video features such as edges and curves. By identifying those features across the various layers, V-Nova says that the amount of residual data required for reconstruction is reduced and allows Perseus Pro to deliver a truly accurate picture, even at lower bitrates.
“P.Link and the Perseus Pro codec that powers it had already revolutionised the contribution space by enabling studio-quality, intra-only HD and Ultra HD at vastly lower bitrates than existing solutions,” explained V-Nova CEO and co-founder Guido Meardi. “With our machine-learning advances, we’re now able to deliver visually lossless UHD at just 140 Mbps, which is comparable with what many operators are currently using to deliver HD.”
Oggi alle 18:25
