V-Nova’s Perseus reincarnated as SMPTE VC-6

Ever since its launch in April 2015 as an alternative codec incorporating a different approach to traditional technologies in the MPEG mold, V-Nova has suffered from two problems. It has struggled to gain significant customers beyond its launch partner Sky Italia and been demonized as an outsider by many in the mainstream codec community. In response, the Perseus technology has been expanded with the help of machine learning and been recast more as an enhancement or accelerator for existing codecs rather than a competitor to them.

As part of this process, V-Nova partitioned its product into two, Perseus Pro for production and contribution encoding, and Perseus Plus for enhancing the compression and computational performance of existing codecs such as AVC/H.264, HEVC, VP9 and in the future AV1.

For Perseus Pro, now rechristened PPro, V-Nova developed machine learning algorithms based on neural network convolution, which is particularly well suited to analysis of visual images. Originally, Perseus was oblivious to the structure of the underlying image, operating almost purely on mathematical redundancy within the digitized data so that it could equally well compress any set of binary information, not just video. But V-Nova saw the potential and indeed necessity for achieving further gains in compression efficiency by taking account of the spatial data structure within images.

The machine learning algorithms dovetail with the hierarchical multilayer structuring of the images performed by Perseus, manipulating connections and their weights to differentiate finely between objects at different scales. This enables objects, or rather their boundaries, to be identified with the smallest possible residuals, in essence number of bits. The benefits are better overall compression compared to comparable production and imaging formats and in particular enhanced rendering at multiple scales, such that effectively lossless video can be reconstructed after decoding given sufficiently high bit rates. At lower bit rates, the codec can still avoid blocky artefacts after reconstruction to yield video that appears almost visually lossless.

Perseus Plus then applies some the same algorithms to enhance those existing codecs. In this guise, the product is overcoming the stigma of failing to win over big customers, since it can be employed readily as a software enhancement to other codecs already installed.

The latest development however features PPro, which has provided the essence of the new SMPTE standard ST-2117, or VC-6 as it will be more widely known. This confirms the status of the upgraded Perseus as an accepted mainstream standard and also its position as an established codec for contribution where it has gained more traction from the outset. Indeed, Sky Italia initially employed Perseus only for video contribution.

SMPTE was attracted by the hierarchical representation of compressed data enabled by PPro, which allows decoders to recreate uncompressed imagery flexibility, with higher resolutions overlaid on the underlying intra-only frames. The use of machine learning then allows content complexity to be taken into account, helping meet SMPTE’s objective of enabling greater control over the resulting quality after decoding, according to the bandwidth available. The same algorithms can also be recast for object-based image search and indexing.

The result is an alternative contribution codec to the long-established JPEG 2000 and MPEG-1, capable of achieving visually lossless compression at lower bit rates down to 150 Mbps. While technically, or mathematically, lossless compression recreates the video identically after decoding, visually lossless merely reconstructs images that appear the same as the original to the eye. It is subjective and therefore has to be defined by some measure such as SSIM (Structural Similarity Index) for comparing two video sources on an agreed scale.

The SMPTE work is further evidence of V-Nova’s conversion from upstart to mainstream player, following its joining the Advanced Television Systems Committee (ATSC) in June 2019 to participate in that group’s work on enhancing existing video codecs.