A practical guide to video compression standards for the telecom industry
The abbreviations AVC, HEVC, and VVC come up constantly in technical discussions. But what do they actually mean in practice? Why can’t we simply choose one standard and stick with it forever? And when does it really make sense to consider switching to VVC?
Let’s break it down without getting lost in unnecessary theory.
Why Do We Need New Standards?
The reason is simple: hardware is becoming more powerful, video resolutions are increasing, frame rates are getting higher, and the number of bits per pixel is growing as well. All of this has to be transmitted and stored while keeping bitrates within reasonable limits.
Each new standard is developed with a clear goal: it should be at least 50% more efficient than its predecessor. At the same time, decoding should not be more than twice as complex, while encoding should not be more than ten times as complex.
Three Generations of Video Standards: What Has Changed?
AVC (H.264) is still the workhorse of the industry. It introduced a number of key tools that became fundamental to modern video compression, including angular intra prediction and Context-Adaptive Binary Arithmetic Coding (CABAC). Seven angular prediction modes and square transform blocks may look modest by today’s standards, but they represented a major step forward at the time.
HEVC (H.265) was the next generation and brought a significant improvement in compression efficiency. The number of angular prediction modes increased from 7 to 33. Tiles were introduced to enable parallel encoding of multiple image regions. The size of the basic processing block also increased.
VVC (H.266) is the latest standard, finalized in recent years. It was developed as a direct continuation of HEVC, with a clear goal: to significantly reduce bitrate once again while maintaining the same visual quality. The number of angular prediction modes doubled compared with HEVC. According to measurements, VVC is approximately 40% more efficient than HEVC in terms of PSNR.
What Does VVC Bring to the Table?
VVC inherits all the tools available in HEVC and adds a range of new ones.
More flexible block partitioning. HEVC used only a quadtree structure, which recursively divides a block into four parts. VVC adds binary and ternary partitioning, allowing a block to be split into two or three parts. This makes it possible to adapt more precisely to image content.
Affine motion compensation. Before VVC, motion prediction was limited to a single approach based on so-called translational motion. In reality, motion can be much more complex: blocks can rotate, zoom in or out, and undergo shear distortion. To support these more complex types of motion, VVC introduced affine motion compensation.
Matrix-based intra prediction. This is a fundamentally new type of intra prediction based on machine learning. Unlike the other tools in the standard, it does not have a strict mathematical foundation—it simply works well in practice.
Larger basic processing block. The size of the Coding Tree Unit (CTU), the basic processing block, was increased to 128×128 pixels. This reduces signaling overhead when working with high-resolution video.
Subpictures. A new data structure that makes it possible to encode multiple independent images within a single frame. This is particularly useful for 360° video, one of the key use cases considered during the development of VVC.
Screen content coding profile. VVC includes support for encoding content captured from a screen in its main profile.
Why Isn’t VVC Everywhere Yet?
If VVC is so much better, why isn’t it used everywhere? The short answer is: hardware support and computational complexity.
VVC encoding takes approximately 20 times longer than HEVC encoding. HEVC already requires significant computational resources—and VVC can require roughly 20 times more. This is a serious obstacle for real-time video encoding.
There are still very few hardware encoders and decoders that support VVC, with only a handful of models currently available on the market. The history of the industry also shows that early hardware implementations often support only a subset of a standard’s capabilities.
For example, when Intel released its first hardware HEVC encoder, it was technically an HEVC implementation, but in terms of the tools actually used, it was closer to AVC—with just seven angular prediction modes instead of 33. A more complete implementation came later in subsequent generations.
When Should You Switch to VVC? Practical Considerations
VVC can significantly improve compression efficiency compared with HEVC. Based on the estimates presented in the talk, the efficiency gain is around 40%, and under otherwise equal conditions, switching to VVC can substantially reduce bitrate.
However, compression efficiency is not the only factor to consider when choosing a new codec. Computational complexity and hardware support are equally important.
VVC is worth considering if:
compression efficiency and bitrate reduction are important to you;
you work with high-resolution video—in particular, 8K was one of the resolutions given special attention during VVC development;
you can afford the higher computational complexity of encoding, for example, in scenarios where encoding speed is not the primary constraint;
your users’ devices and software support VVC decoding.
You should be more cautious about switching if:
you need real-time encoding and have limited computational resources;
your audience uses devices without hardware or software support for VVC;
you require a complete hardware implementation of all the standard’s capabilities.
Ultimately, the right choice depends on the specific use case. VVC offers higher compression efficiency, but that comes at the cost of greater encoding complexity and still-limited hardware support. Before making the switch, it is therefore important to weigh the potential bitrate savings against the computational costs and the capabilities of the devices that will be used to decode the stream.
Top comments (0)