Technically speaking, a video codec is nothing more than an algorithm. Its main task is to reduce huge volumes of data in a video recording so the data can be stored, copied, streamed, and played back efficiently. A video codec saves storage space on your mass media and reduces the bandwidth required for streaming.

The freely-licensed AOMedia Video 1 (AV1) [1] and the patented (commercial) Versatile Video Coding (VVC, also known as H.266) [2] are two modern video codecs that lead the field in terms in efficiency and speed. Both promise up to 50 percent better compression than their predecessors. These codecs not only differ in terms of licensing rights, but also in terms of technology and strategic orientation. This article highlights the most important differences between AV1 and VVC with a view to the licensing models, performance, and suitability for specific applications, using benchmarks to clarify who comes out on top.

AV1 and VVC are both primarily lossy compression methods: They reduce a video's file size by permanently masking or removing certain data (usually data that is of little importance to human perception). Because of this, they can maintain a very high level of quality from the user's point of view, with compression artifacts usually remaining inconspicuous. Both codecs optionally support a lossless mode in which there is no loss of quality, although this is rarely used in standard operation.

AVI

AV1 was launched in 2018 by the Alliance for Open Media (AOMedia), an industry consortium that includes Google, Cisco, Amazon, Netflix, Microsoft, Mozilla Corporation, and Intel. AOMedia's goal is to develop and promote open, royalty-free standards and technologies for media transmission in the areas of audio and video. The challenge is to ensure high video quality with the lowest possible bandwidth and hardware requirements so that the format can be used on all devices and with a wide variety of network connections.

The AV1 codec's official reference implementation is the libaom library, which is optimized for completeness rather than speed. It is used in Google Chrome, to cite just one example. As an alternative to libaom, the dav1d library, developed by VideoLAN (VLC) and FFmpeg with support from the AOMedia, aims to provide the most powerful decoding possible, but does not include an encoder. Applications such as Firefox, VLC, and the Netflix Android app use dav1d because the library is particularly efficient and requires less memory than libaom. Because no license fees are required for AV1, the codec is particularly interesting for web platforms with video offerings, such as YouTube, Netflix, and Amazon, but also for hardware manufacturers.

A further development of Google's VP9 (introduced in 2013), AV1 also uses techniques from Mozilla's Daala and Cisco's Thor codecs. AV1 is slightly slower than VP9 for encoding due to its use of more complex algorithms, especially on older hardware, but it does offer higher compression rates and superior output quality at identical bitrates. For many streaming applications, this is a big leap forward, helping to save bandwidth and reduce storage costs. Video data encoded with AV1 uses the MP4 or MKV container format, which forms the basis for WebM in combination with the Opus audio format.

VVC

VVC (H.266) is the official successor to HEVC (H.265) and was developed in 2020 by the Joint Video Experts Team under the auspices of ITU and MPEG. Organizations such as Fraunhofer Heinrich Hertz Institute (HHI), Apple, Huawei, Qualcomm, and Ericsson back VVC. To legally use the patented codec, license fees are required. VVC is managed by several patent pools such as MPEG LA and VVC Advance. Licensing costs depend on the specific usage context, which hinders VVC's widespread adoption, particularly in open source and web applications.

VVC is no faster than its predecessor HEVC when encoding, but it is far more efficient in terms of compression. The codec reduces the bitrate by about 50 percent compared to its predecessor while maintaining the same quality. However, this does require significantly more computing power and takes about five times as long to encode. Compared to AV1, it offers better compression efficiency, but AV1 encodes and decodes content far faster.

Both codecs use state-of-the-art methods such as partitioned block coding, intra- and inter-prediction, transformation methods, entropy coding, and advanced motion compensation. However, VVC offers even more sophisticated algorithms for scenes with very high resolution, such as 8K or volumetric videos.

Distribution

AV1 is already widely used today. Major platforms such as YouTube, Netflix, and Facebook deliver content with AV1 if supported by end devices. Browsers such as Chrome, Firefox, and Edge already offer AV1 support. Modern GPUs from Intel, AMD, and NVIDIA also support hardware-based AV1 decoding. VVC is still in its infancy, but it is increasingly making a name for itself. The first software decoders, such as the VVdeC [3] from the Fraunhofer HHI, can already be used on standard hardware with four to eight CPU cores and are implemented in software, such as the Elmedia Player [4] for macOS.

Newer Intel processors, starting with the "Lunar Lake" series, can decode VVC [5], while Sharp is developing a real-time decoder for 8K VVC videos [6]. You can also expect hardware decoders to be installed in some upcoming high-end smartphones from Huawei's Mate series with the latest Kirin processors. Mediatek offers various software-on-chips (SoCs) that support VVC in its Pentonic series, which are used in 4K smart TVs by Sony, Panasonic, Philips, Samsung, and Xiaomi, among others.