VVC vs. MPEG-5 EVC / LCEVC: which standard will mark the future of broadcasting?
In this Tribune, Dr. David Guillermo Fernández Herrera, CTO of AI VIDEX, takes an in-depth look at VVC and MPEG-5 EVC/LCEVC video compression formats, which are great candidates to become a standard in the broadcast industry in the short and medium term.
Over the past few decades, we have experienced an incredible advancement in the field of multimedia applications. Among the various factors involved, it is worth highlighting the contribution of the Video compression standards. Since the late 1980s, a multitude of standards have been developed, from H.261 to the recent VVC (Versatile Video Coding), always looking for high data compression rates without reducing visual quality.
This objective has allowed that, approximately 10 years, a new standard is released that provides a 50% reduction in bandwidth required over its predecessor. This has been the case of H.265/HEVC (High Efficiency Video Coding, released in 2013) against H.264/AVC (Advanced Video Coding, 2003).
In 2020, the first version of the standard was published VVC and two other new standards within the family MPEG-5, MPEG-5 EVC (Essential Video Encoding) and MPEG-5 LCEVC (Low Complexity Enhancement Video Coding).
Discovering VVC
VVC Promises reduce by 50% the generated data rate (bitstream) with respect to H.265. However, this improvement comes at the expense of considerably increasing the Algorithm complexity and consequently, require greater computational resources, higher power consumption and, therefore, higher economic cost to meet the requirements that allow the real-time encoding/decoding of high (HD) and ultra-high definition (UHD) signals. This is something we have become accustomed to with the release of each new standard, and VVC has been no exception, as it features a large number of New techniques that add great complexity to the design of an encoder or decoder.
Both the rate of compression as the Visual quality of the video depend directly on the implementation made in the encoder, of what techniques have been used and how they have been Implemented.
It is very important to highlight that video compression standards define the syntax of the Bitstream (how to construct it) and the method to follow to perform the decoding (how to reconstruct a video from the Bitstream), but do not describe the implementation of the encoder. This means that the standard indicates, for example, how to convert a motion vector to a bitstream, but does not define how the encoder obtains the value of said vector by looking for the movement between the different images that make up the video. Both the rate of compression as the Visual quality of the video depend directly on the implementation made in the encoder, of what techniques have been used and how they have been Implemented. Therefore, the capabilities of an encoder can vary significantly from one implementation to another.
In the first implementations of H.265 encoders, it could be corroborated that, when compared with mature implementations of H.264 encoders using configurations that allow the same amount of computational resources (CPU load, memory usage, etc.), better results were obtained in H.264. Currently, as there are already sufficiently mature H.265 implementations, the situation has changed. The purpose of this commentary is to Alert on encoder deployments that arise after the release of a new standard. Some are conceived as mere marketing products that include simple implementations of new techniques seeking to reach the market as soon as possible, but finally fall under their own weight when detailed comparisons are made with mature coders based on previous standards. Implementations made from an exhaustive study of the standard and the state of the art, and which have been optimized in strict test environments, are those that they really exploit the contributions of each new standard.
Steps forward with VVC in the UHD world
Among the contributions within VVC, it is worth highlighting the use of large encoding units (128×128 pixels), which allow for a huge reduction in the Bandwidth required when applied in areas of Homogeneous textures in the resolution UHD. Another very interesting contribution is the use of Multiple Partition Types, which allows the image to be divided into encoding units that adapt to the content, perfectly delimiting the different objects of the image. This level of division opens the door to the use of artificial intelligence algorithms in video compression, such as segmentation, where each pixel of the image is classified individually. By means of segmentation, regions, individuals or objects of interest can be detected and delimited in different classes within each image and different compression techniques can be applied according to the level of visual quality that is desired for each class. In other words, to allocate More bits to regions of interest to improve its visual quality based on a previous analysis of the image.
VVC provides the Tools Needed to optimize the distribution of UHD content and there are already commercial implementations that make it possible; but such solutions use Lots of computational resources and suppose a Significant investment.
These tools are just a small introduction to all those used by VVC and that, when applied correctly, can represent a very important qualitative leap. As more articles are published on how to efficiently use these techniques and greater efforts are dedicated to optimizing the different algorithms, the results will begin to be obtained VVC Deployments that allow them to improve the performance of their predecessors at a reasonable computational cost. VVC provides the necessary tools to optimize the distribution of content in UHD and there are already commercial implementations that make it possible; but such solutions use a large Amount of computational resources and suppose a Significant investment.
MPEG5-EVC/LCEVC: Natural Evolution?
MPEG-5 EVC it has the advantage of having a basic royalty-free profile, thus solving a weakness of H.265, where license management has become a very cumbersome process. But within the MPEG-5 family, it stands out MPEG-5 LCEVC.
This new standard is based on adding Additional layers of quality improvement on a base layer obtained using other compression standards. This concept is a Big Leap compared to its predecessors, as it allows teams to continue to make use of Hardware acceleration (based on H.264, H.265, VP9, etc.) to process the base layer and, by software, improve its performance thanks to additional layers. On the set-top box side, if MPEG-5 LCEVC is not supported, only the base layer will be decoded allowing compatibility with existing infrastructures and devices.
Therefore, MPEG-5 LCEVC introduces the Compression Enhancement Data Flow Concept, Reduces processing complexity and is backward compatible; all without introducing additional latency into the system.
Adopting VVC and MPEG-5 LCEVC
Although three years have passed since the release of the first version of the standard VVC, there is currently no hardware support in any device for VVC, which slows its adoption and allows us to predict that its expansion will not accelerate until 2026 or 2027. Even without hardware support, we are seeing a number of indicators in the market that indicate its potential. One of these indicators is the adoption of VVC in application standards such as DVB (adopted VVC as Next Generation Codec, in 2022), SBTVD (for its base layer, in 2021), SCTE (included it among its standards in 2023) and its inclusion in ATSC 3.0. The effort dedicated to the implementation of VVC solutions by companies such as Qualcomm, Huawei, Hikvision, MainConcept, Ateme or Bitmovin, could be another indication of its future wide adoption.
MPEG5-LCEVC is a palpable reality, its implementation is currently feasible, and by adding Software Layers both in encoding and decoding, it provides a significant improvement in the performance of any system based on previous standards.
Due to the lack of implementations Open source VVC mature and the resources allocated by commercial coders (8 times more than H.264 and 4 times more than HEVC), MPEG-5 LCEVC It is emerging as a very interesting option to improve the performance of existing systems based on previous standards. Of course, in the case of content or multimedia service providers, where the possibility of dedicating Powerful servers to encoding Don't be a problem, you shouldn't lose sight of VVC by the Significant improvements that it contributes to compression efficiency and by the indicators on its future adoption that we have discussed.
As deployments progress, lower-cost devices will begin to appear where VVC can run. The same was true of its predecessors: early H.265 implementations could only run on powerful servers; but, over the years, optimized implementations came running on other types of devices such as FPGAs, GPUs, and eventually ASICs. With this, its use was democratized.
Although VVC currently it does not have devices that perform decoding by hardware acceleration (mobiles, set-top boxes, GPUs in computers, etc.), it is emerging as the Great Candidate for decrease the bandwidth required to distribute content UHD in the coming years.
MPEG5-LCEVC is a palpable reality, its implementation is currently feasible, and by adding software layers in both encoding and decoding, it provides a significant improvement in the performance of any system based on previous standards. Thanks to the compatibility it offers, it can be deployed using the current infrastructure and without the need to create a parallel data flow, which means a significant cost reduction compared to the implementation of a standard that requires the application of flows due to incompatibility problems. However, although VVC currently it does not have devices that perform decoding by hardware acceleration (mobiles, set-top boxes, GPUs in computers, etc.), it is emerging as the great candidate to decrease the bandwidth required to distribute UHD content in the coming years.
In the end, it will be the bets of broadcasters, manufacturers and suppliers, as well as specific needs in the increasingly broad concept Broadcast, who will define the winner in a battle with Two very powerful technologies that solve many of the challenges involved in streaming content HD and UHD.
Dr. David Guillermo Fernández Herrera
CTO of AI VIDEX
Did you like this article?
Subscribe to us RSS feed And you will not miss anything.