By 2020, it is predicted that video will account for roughly 80% of all Internet traffic, 20% of which will be Ultra HD. AV1 the codec designed for delivering OTT content that is backed by the likes of Amazon, Netflix and Google, threatens to challenge the current dominance of HEVC. Meanwhile other challengers have emerged. V-NOVA launched its compression technology in April 2015, claiming “2x–3x average compression gains, at all quality levels, under practical real-time operating scenarios versus H.264 and HEVC.” Since then, V-Nova has led the evolution of the video infrastructure industry, adding machine learning capabilities at the core of its flagship Perseus codec.
Services like Netflix stream 4K video at over 15 Mbps and a few services distribute live 4K content at substantially higher rates - typically 25 – 35 Mbps. PERSEUS Plus can cut 4K bitrates in half. Fabio Murra, SVP Product & Marketing at V-NOVA, explains how their technology can make a difference.
Digital Studio: How does V-Nova’s PERSEUS Plus codec achieve the compression level needed for UHD streaming at 8 Mbps?
PERSEUS Plus utilizes a unique hierarchical approach to video compression that works with other industry-standard codecs to add an additional layer of quality and efficiency. In a UHD stream, PERSEUS Plus H.264 creates a lower resolution H.264 encode and then the PERSEUS Plus enhancement layer enables a highly efficient up-sample to UHD whilst adding in any residual detail required.
In this way, by working in tandem with your current underlying video codec and streaming formats (e.g. HLS with H.264), PERSEUS Plus can provide the step-change in video quality and efficiency that HEVC and AV1 are yet to deliver. Furthermore, with PERSEUS Plus operators do not need to worry about device incompatibility or the vast costs associated with upgrading components across an entire delivery workflow.
DS: Aside from the growth in UHD distribution, what other trends in the industry are driving demand for advanced video compression?
Yes, there is growing interest for live 1080p60 and 4Kp60, particularly from sports broadcasters and the rapid growth of hyperscale eSports services. For these high-action use cases, using 25 or 30fps streaming fails to deliver a smooth enough representation of the content. Today, consumers are increasingly aware of these types of poor video quality experiences. Most devices support 1080p60 playback but cannot reliably receive the 6 Mbps or more that traditional codecs require for this, so there’s a huge opportunity to address this issue with better compression.
DS: How important is Artificial Intelligence for achieving content aware encoding in UHD streaming and video content delivery?
There is great potential for Artificial Intelligence (AI) to help make significant advances in video compression. Some simple applications of AI have already been tried by companies; for example, applying AI to evaluate and choose between different encoding parameters.
While this only provides small incremental benefits, there are more fundamental applications at the core of a codec which can be realized. For example, at V-Nova we have invested quite heavily in AI and in PERSEUS Pro (our contribution, imaging and production codec) we have deployed convolutional neural networks (CNNs) to enable the codec to learn about common image features. This means that when we up-sample from our lower resolution base encode we are able to more precisely reconstruct the higher resolution image, saving in the region of 25 to 30 percent in bitrate.
DS: Has 4K and HDR seen the kind of adoption expected so far?
To date, the adoption is probably what most would have expected. Initially, production costs were very challenging which restricted its use to global media players like Netflix. The steadily falling costs of UHD production and the rapid increase in market penetration of 4K TVs in the home is driving increased demand across the ecosystem for 4K delivery, but it is taking time for these dynamics to take hold.
However, with the move to 4K delivery incurring substantial increases in bandwidth requirements, the video compression used becomes a major barrier to the rollout of the services consumers want to see. These trends have driven a tremendous increase in demand for our technology this year across the entire delivery ecosystem, from contribution through to distribution to viewers.
DS: What trends did you notice at the BroadcastAsia event this year?
Asia is currently experiencing two extreme market sectors when it comes to advancing technology; i.e. Japan and Korea are pushing 4K capabilities and beyond, but the rest of Asia is still trying to make the move from SD to HD.
Global online content providers are leading the way in 4K, but regional operators don’t have the scale to absorb the costs involved. In order for 4K to be commercially viable, it needs much lower bitrates; for example, it needs to be closer to today’s HD costs as there are currently no additional revenue opportunities from providing the service.
DS: How was your experience partnering with MEASAT and NovelSat to demonstrate UHD primary distribution at BroadcastAsia?
It was tremendous to see such a great response to what we built with MEASAT and NovelSat to provide such a significant step forward for the viability of UHD primary distribution. In the past, delivering live UHD at the quality required at just 8Mbps has been impossible with other solutions. Now, we are able to make 4K content distribution significantly more cost-effective and viable for broadcasters needing to contribute their channels to a variety of operators.
DS: There has been a lot of talk of the battle between HEVC and AV1 recently, but what’s your take?
HEVC adoption has been undermined by onerous and unclear royalties, and support in devices and browsers is still very limited. This explains why Google, Amazon and Netflix are promoting a royalty-free alternative. But AV1 also needs new hardware, which will take many years. Neither HEVC nor AV1 can address the billions of devices in use today – and for the next five to 10 years. This means that operators who want to use these codecs have to duplicate workflows and deliver multiple ABR ladders.
In addition, both HEVC (and especially AV1), are extremely processing hungry, which makes them costly for the live and time-shifted video that represent the majority of what people watch today. That is why we developed PERSEUS Plus AVC/h.264, which delivers better-than-HEVC compression, is faster to encode than AVC/h.264, and can be played on all existing consumer devices in hardware without reducing battery life. It is also plug-in free in any browser. It is effectively “royalty already paid for”, since it requires no additional IP hardware blocks in devices. Additionally, PERSEUS Plus is codec agnostic. When HEVC or AV1 are decodable, PERSEUS Plus HEVC/AV1 can deliver substantial improvements relative to native HEVC/AV1.
Winning the Codec Battle
Given the price sensitive nature of low-cost mobile devices, hardware manufacturers cannot afford to pay millions of dollars in royalties to enable proprietary formats on devices, because it’s too much of a cost. That’s why the industry has reacted strongly to the need for royalty-free codecs. Development of AV1 was a response to that need, leading to the current faceoff between AV1 and HEVC to become a standard codec for Ultra HD video.
V-Nova’s PERSEUS is already able to achieve that without necessarily requiring a change in hardware. But until it becomes widely accessable across devices and platforms, it’s not going to be adopted by a lot of people. At the same time, the mission to drive higher quality video and efficiency in compression continues for all players in the streaming video ecosystem.
Competing with compression codecs backed by the biggest stakeholders in the video ecosystem is not an easy task. V-Nova’s Perseus Pro technology is impressive and a proven cost saver, yet the proprietary codec faces a challenge to gather momentum.