V-Nova LCEVC consists of a comprehensive SDK and an NDK for enhancing any encoding and decoding workflow with MPEG-5 Part 2 LCEVC. LCEVC provides a boost to the compression efficiency of any existing or future video codec, enabling higher quality at up to 40% lower bitrates while accelerating encoding and reducing compute costs by up to 4x.
Whatever encoding and playback platforms or devices your product or service employs, V-Nova LCEVC provides a broad array of options to integrate at any level of your software stack or to work with solutions from third-party encoder and player vendors.
LCEVC does not mean ‘software only’. For integration in TVs and STBs, V-Nova has created the LCEVC NDK (Native Development Kit) aimed at supporting the development of on-chip capabilities for the encoding and decoding of LCEVC.
Deploying LCEVC on the chipset using the V-Nova LCEVC NDK guarantees:
- improved content security
- increased decoding performance
- reduced power consumption
V-Nova LCEVC NDK is a collection of tools to Video SOC vendors and ASIC designers to accelerate the availability of LCEVC on their roadmap. The toolkit includes optimized libraries that can be used at driver level as well as fully functional and compliant ASIC IP blocks. These blocks are made available as RTL code with the relevant documentation and a group of Integration Services resources at hand to customize them for a given chipset architecture.
Wherever your encoding runs, from cloud infrastructure to on-prem or personal devices, the V-Nova LCEVC SDK has the broad coverage needed to deploy optimised LCEVC-enhancement. Crucially, LCEVC is a software update to your existing workflows and is transparent to other systems like packaging, ad-insertion and CDN.
Integration and testing is typically a matter of days rather than weeks. If you do not have the resources available internally we have dedicated integration services to handle everything for you.
The LCEVC SDK uses a plug-in system via our Encoder Integration Layer (EIL) to enable the interface between the base encoder and the LCEVC enhancement. Over 20 encoders are already supported including x264, x265, libaom, libvpx, NVenc, MainConcept H.264/HEVC and more are being added all the time.
Up to 4x more throughput
LCEVC typically uses the base encoder at 1/4 resolution and then employs the low-complexity optimised toolset from V-Nova LCEVC. This accelerates encoding by up to 4x for codecs like AV1 or VVC and up to 2x for AVC/H.264.
The LCEVC SDK includes a decoder integration layer (DIL) which provides seamless integration into players on a host of device types whilst including optimizations that harness the full capabilities of different CPU, GPU and SoC architectures. And thanks to a range of reference integrations for widely used players like ExoPlayer, VLCKit and hls.js among others, deployment is quick and easy.
The DIL and our reference implementations are available across Android, iOS, Windows, MAC and HTML5 meaning that the vast majority of devices can be quickly updated in software to enable LCEVC-enhancement decoding.
The LCEVC decoder integration layer leverages standard GPU resources on devices to decode the LCEVC enhancement whilst power consumption for the base is reduced, ensuring battery life for media consumption is not impacted.
LCEVC is backwards compatible meaning that decoders that have not been updated for LCEVC can still play the base video. For this reason, the same streams can address both legacy and updated devices and the rollout of LCEVC across multiple screens can be phased in as required.
Kineton performed over 5,000 individual tests of V-Nova LCEVC on Android and iOS video players. The 10-week testing period recreated the end-user experience and focused on four key areas of measurement: frames-per-second playback, dropped frames, adaptive bitrate performance and power consumption.
Results revealed over 99% success rates (99.8% for iOS and 99.7% for Android). V-Nova LCEVC has now been validated for use in software for mobile video applications.
MPEG-5 LCEVC works by encoding the source video at a lower resolution (typically 1/4 resolution) using any existing codec (the base codec), corrections are applied to that base, a content-intelligent upsampling is conducted and a final layer of spatial details are applied. The LCEVC enhancement process is low-complexity by design providing substantial compute savings compared to using the base encoder in full resolution.
V-Nova LCEVC incorporates a range of unique enhancement tools and optimisations that operate within the framework of the MPEG-5 LCEVC standard to deliver a step-change in video compression performance. V-Nova’s LCEVC implementation leverages the principles of hierarchical image representation and massively parallel processing and is the culmination of over 7 years of innovation and development.
For more information on the MPEG-5 Part 2 LCEVC specification visit www.lcevc.org.
There is a range of ways to test LCEVC-enhanced encoding and playback for your service or application. Encode on the V-Nova platform and playback in our demo application or test locally with FFmpeg and our reference implementation players.
This report illustrates the benefits of the MPEG-5 LCEVC enhancement applied to NVIDIA NVENC H.264 and HEVC, using an ultra low latency and GPU processing configuration relevant for use cases such as Cloud Gaming, split-compute rendering or XR streaming.
The graphs below illustrate the significant increase in compression efficiency achieved by applying LCEVC enhancement to UHD content based on the VMAF metric. For both x264 and x265 encoders, the same quality can be achieved at 45% lower bitrates in the critical range needed to deliver satisfactory visual quality (typically VMAF scores between 90 and 95).
Tests have been conducted across a vast range of 1080p content. Here is an example of how LCEVC enhancement affects the performance of x264 and x265, using CRF rate control. LCEVC-enhanced x264 achieves a compression efficiency performance that is equivalent, or better than, x265. These gains are achieved across the full range of bitrates at which these codecs are employed.
LCEVC has been thoroughly tried and tested by independent organisations and individuals. Jan Ozer, SPIE and MPEG itself are just a few sources of the reports presented below.