H.266, also called Versatile Video Coding (VVC), is the international video compression standard finalized in 2020 by the ITU-T and ISO/IEC, published as ITU-T H.266 and MPEG-I Part 3. Among video codecs, it is the direct successor to H.265 (HEVC) and targets about 50 percent lower bitrate at the same perceived quality, measured mainly on 4K and 8K content.
The word versatile reflects its scope: beyond conventional video, VVC includes tools for screen content, 360-degree video, HDR, adaptive resolution and layered coding in a single specification. The price for this efficiency is much higher encoder complexity and, so far, limited hardware decoding support in consumer devices.
How H.266 compresses video
H.266 keeps the hybrid block-based architecture used since MPEG-2: each frame is split into blocks, every block is predicted from neighboring pixels or other frames, the residual is transformed and quantized, and the result is entropy coded with CABAC. The gains come from refining nearly every stage rather than from a new paradigm.
The most visible change is the partitioning. VVC uses coding tree units of up to 128×128 pixels and a multi-type tree that combines quadtree splits with binary and ternary splits, so blocks can be rectangular and follow object edges closely. Intra prediction grows to 67 directional modes, and inter prediction adds affine motion compensation for zoom and rotation, decoder-side motion refinement and bi-directional optical flow.
In-loop filtering is also stronger. After the familiar deblocking filter and sample adaptive offset, VVC applies an adaptive loop filter (ALF) and luma mapping with chroma scaling (LMCS), which redistributes code values to improve quality, particularly for HDR material.
| Coding stage | H.265 (HEVC) | H.266 (VVC) |
|---|---|---|
| Largest block | 64×64 coding tree unit | 128×128 coding tree unit |
| Block partitioning | Quadtree only | Quadtree plus binary and ternary splits |
| Intra prediction | 35 modes | 67 modes plus matrix-based prediction |
| Inter prediction | Translational motion | Adds affine motion and optical flow refinement |
| Transforms | DCT and DST, square | Multiple transform selection, rectangular blocks |
| Loop filters | Deblocking and SAO | Deblocking, SAO, ALF and LMCS |
Profiles, levels and special features
Like earlier standards, H.266 defines profiles that fix the allowed tools and levels that cap resolution, frame rate and bitrate. The most widely targeted profile is Main 10, which covers 4:2:0 video at up to 10 bits per sample and therefore suits both SDR and HDR delivery.
| Profile | Chroma and bit depth | Typical use |
|---|---|---|
| Main 10 | 4:2:0 or monochrome, up to 10-bit | Streaming and broadcast delivery |
| Main 10 4:4:4 | Up to 4:4:4, up to 10-bit | Screen content and graphics |
| Multilayer Main 10 | 4:2:0, up to 10-bit, several layers | Scalable and multiview services |
| Main 10 Still Picture | 4:2:0, up to 10-bit, single frame | Still image coding |
Several features address delivery problems directly. Reference picture resampling lets the stream change resolution without an intra frame, which helps adaptive streaming. Subpictures allow independent regions to be extracted, useful for viewport-dependent 360-degree video. Screen content tools such as intra block copy and palette mode compress text and user interfaces far better than camera-oriented tools.
How H.266 compares with other codecs
H.266 competes mainly with H.265 and with the royalty-free AV1. Published test results consistently show VVC ahead in pure compression efficiency, while AV1 leads in browser and hardware support and licensing simplicity. The exact percentage savings depend strongly on content, resolution and encoder settings, so any single figure should be read as an average.
| Codec | Relative efficiency | Licensing model | Decoder availability |
|---|---|---|---|
| H.264 | Baseline reference | Patent pool, widely settled | Universal |
| H.265 | About 50 percent better than H.264 | Several patent pools | Most modern devices |
| AV1 | Similar to or slightly better than H.265 | Royalty-free by design | Browsers and recent chips |
| H.266 | About 50 percent better than H.265 | Patent pools still forming | Limited, growing slowly |
Compared with intra-frame codecs such as ProRes, H.266 is a long-GOP delivery codec: it relies on inter prediction across many frames, so it is unsuitable as an editing format and is used at the end of the chain, after mastering.
Encoding and deployment in practice
The reference encoder, VTM, is a research tool and runs far slower than real time. Practical work uses optimized encoders such as the open-source VVenC and decoder VVdeC, which offer speed presets that trade compression for time. Even fast presets demand several times the computation of an HEVC encode at comparable settings.
Deployment is strongest where bandwidth is expensive and devices are controlled: broadcast specifications such as DVB and the Brazilian TV 3.0 system have adopted VVC, and some 8K and set-top box chips include decoders. General web streaming is slower to follow because mainstream browsers do not decode VVC natively.
| Situation | Recommendation | Reason |
|---|---|---|
| 8K or premium 4K broadcast | Strong candidate | Bandwidth savings outweigh encode cost |
| Managed IPTV with known set-top boxes | Good candidate | Hardware decoding can be specified |
| Open web streaming to browsers | Use AV1 or H.265 instead | No native browser decoding |
| Live low-latency encoding on modest hardware | Usually avoid | Encoder complexity is too high |
Limitations and common misconceptions
- Fifty percent is an average, not a guarantee. Savings shrink at low resolutions and with fast encoder presets.
- Higher efficiency means higher cost. Encoding time and energy rise sharply, and decoders also need more processing than for HEVC.
- Licensing remains a real barrier. Multiple patent pools and individual holders make royalty planning harder than for AV1.
- H.266 is not H.265 with a new name. Bitstreams are incompatible, so existing HEVC decoders cannot play VVC files.
- Container support varies. VVC can be carried in MP4 and MPEG-TS, but many players and tools still do not recognize it.
In plain words
H.266 is a newer and smarter way to shrink video files. It studies each picture more carefully than older methods. It describes shapes and movement more precisely. As a result, it needs roughly half the data of H.265 for a picture that looks the same.
The catch is the effort involved. This careful work demands much more computing power. Many phones, televisions and browsers cannot play it yet. Think of it as an excellent packing method that saves space but takes longer to pack. You also need the right tool to unpack it.
A real-world example
A national broadcaster plans a satellite channel in 8K at 50 frames per second with HDR. With H.265, a stable picture would need around 80 to 100 Mbit/s, which uses a large share of a transponder. Engineers test VVC with an optimized encoder and find that comparable quality is reached at roughly 40 to 50 Mbit/s on their test material.
Because the broadcaster supplies certified receivers, it can require hardware VVC decoding in every box. The encoding hardware costs more and draws more power, but the saved capacity allows a second channel on the same transponder. The outcome is that bandwidth savings justify the higher encoding cost in this controlled environment, while the broadcaster keeps H.265 and AV1 for its web and mobile apps.