Your browser is not supported. Please update it.

30 April 2020

Will LCEVC be the prodigal son to rise from codec war ashes?

One of Faultline’s predictions for 2020 was that the codec wars were destined to intensify after an extended period of relative hush. Most industry projections have since been voided, with various arms of the video industry since joining in solidarity to traverse the challenging terrain thrown up by quarantines, which has only served to elevate the demand for advanced compression techniques.

When this all blows over, the major codec funders and patent pools will forget all about unity and commit full throttle to the dirty tactics that have become associated with the codec industry. We have heard stories from the dirty underbelly of R&D, consisting of academic insults, old-guard snobbery, and even outright lies to block deployments. Our revised prediction for this year then is that the codec wars will still reignite, but will give rise to revitalized smaller players who will not merely be cannon fodder in this fight.

We therefore set out to identify some the biggest potential disruptors in the codec scene once lockdowns are eased and some sense of normality resumes. Looking away from the usual suspects like the big-hitters behind AOMedia and HEVC Advance, or the building momentum behind VVC (versatile video codec), instead we turned to a technology called low complexity enhancement video coding (LCEVC) that is tipped for big things.

Also known as MPEG-5 Part 2, LCEVC is a two-layer video coding standard. The base layer is not tied to any specific technology and can be any video codec, while the higher layer is used to extend the capability of an existing video codec. This bi-layer approach brings backwards compatibility to the fore as a core strength of LCEVC, as non-LCEVC compatible media players will just detect and play the base layer stream.

LCEVC allows any existing encoding pipeline, whether running AVC, HEVC, VP9, AV1 and soon VVC, to achieve the same or better quality as a next-generation codec at the same bitrate and at up to 4x lower computational complexity. How AOMedia, the AV1 backing group, will react to LCEVC will be fascinating to observe, given how LCEVC has been found to reduce limitations in the codec, primarily power consumption.

But will LCEVC live up to its potential? If the standard is deployed widely by tier 1 operators, as inside sources are adamant it will, then the ripple effects on the codec wars could be substantial. Indeed, LCEVC has been described as the only fully software-based video coding method that has no negative effects on battery life and system performance.

With the LCEVC standard only recently being frozen, there are minimal test results to compare with no deployment plans finalized. However, we stumbled across results from StreamingMedia’s own encoding expert Jan Ozer, who found that LCEVC cut about 60% of the encoding time over native x264 at the same resolution, when using the content rate factor (CRF) command string (used for VoD encoding).

Test results were based on encoding a 30-second segment of the 60 fps Netflix Meridian and football test clips, producing an average LCEVC encoding time of 283 seconds, compared to 726 seconds for x264. Here, LCEVC was encoding the base H.264 layer at 960×540 and the enhancement layer to 1080p, while x264 was encoding the complete stream to 1080p.

Another test, using the content bitrate (CBR) command string for live content, found that LCEVC cut encoding time by about 52% on average compared to H.264. LCEVC results came in at 80 seconds, while x264 recorded 166 seconds. When Ozer encoded with 8 threads, results showed LCEVC was still faster than x264, but only by about 25%. However, he also notes observation of some severe transient quality drops in the x264 clips when encoding with 8 threads.

“Clearly, two samples are inadequate to draw any firm conclusions, but quality assessment wasn’t the purpose of this article, and having done all the other work, I wanted to quickly gauge the results,” concludes Ozer.

Faultline has it on good merit that major operators are preparing to deploy LCEVC as soon as this summer – and we will report on any additional LCEVC-based test results as soon as they arrive.