H.265, also known as HEVC (High Efficiency Video Coding) and MPEG-H Part 2, is a video compression standard jointly developed by ITU-T and ISO/IEC and first approved in 2013. Among video codecs, it is the direct successor of H.264, designed to deliver comparable visual quality at a substantially lower bitrate, which made practical 4K, HDR and high-quality streaming possible on existing networks.
H.265 achieves its gains through larger and more flexible coding blocks, more precise prediction and improved in-loop filtering. The price is higher computational complexity and a complicated patent licensing situation, which slowed its adoption on the open web.
How H.265 compresses video
H.265 follows the same hybrid block-based coding model as H.264: each frame is divided into blocks, predicted from already decoded data, and only the prediction error is transformed, quantized and entropy coded. The improvements lie in how flexibly these blocks can be shaped and how accurately they are predicted.
The basic unit is the coding tree unit (CTU), which can measure up to 64 × 64 pixels, compared with the fixed 16 × 16 macroblock of H.264. Each CTU is split by a quadtree into coding units down to 8 × 8, so smooth areas such as sky are coded with a few large blocks while detailed areas receive many small ones.
The following table lists the main coding tools and what they contribute.
| Tool | What it does | Benefit |
|---|---|---|
| Coding tree units | Blocks from 64 × 64 down to 8 × 8 | Efficient coding of large flat areas |
| Intra prediction | 35 modes, including 33 directional | More accurate prediction within a frame |
| Merge mode and AMVP | Reuse and predict motion vectors | Lower cost of motion data |
| Transform units | Transforms from 4 × 4 to 32 × 32 | Better energy compaction |
| Sample adaptive offset | In-loop correction of sample values | Fewer ringing and banding artifacts |
| CABAC entropy coding | Context-adaptive arithmetic coding | Compact final bitstream |
Like other long-GOP codecs, H.265 organizes frames into a GOP structure of keyframes and predicted frames. A keyframe is coded independently, while P and B frames refer to other frames, which provides most of the compression gain.
Profiles, levels and HDR support
H.265 defines profiles that specify which tools and sample formats a decoder must support, and levels that limit resolution, frame rate and bitrate. Devices advertise compatibility as a combination such as Main 10 at Level 5.1.
| Profile | Bit depth | Chroma format | Typical use |
|---|---|---|---|
| Main | 8-bit | 4:2:0 | Standard dynamic range streaming |
| Main 10 | Up to 10-bit | 4:2:0 | 4K HDR delivery and broadcast |
| Main 4:2:2 10 | Up to 10-bit | 4:2:2 | Camera recording and contribution |
| Main 12 and 4:4:4 variants | Up to 12-bit or more | Up to 4:4:4 | Professional and screen content |
Main 10 is the key profile for HDR, because 10-bit coding prevents banding in the extended brightness range of HDR10, HLG and Dolby Vision. Levels up to 6.2 cover 8K resolution at high frame rates.
H.265 compared with other video codecs
H.265 occupies the middle ground between the universally compatible H.264 and the royalty-free AV1. Efficiency gains always depend on content, encoder settings and the quality metric used, so the figures below are typical rather than fixed.
| Codec | Released | Bitrate for similar quality | Licensing | Main strength |
|---|---|---|---|---|
| H.264 | 2003 | Reference | Patent pool royalties | Universal compatibility |
| VP9 | 2013 | Roughly similar to H.265 | Royalty-free | Web streaming support |
| H.265 | 2013 | About 25 to 50 percent lower | Several patent pools | Hardware support for 4K HDR |
| AV1 | 2018 | Often 20 to 30 percent below H.265 | Royalty-free | Efficiency for large platforms |
| ProRes and DNxHR | 2007 and 2014 | Many times higher | Proprietary intermediates | Fast, frame-accurate editing |
Intermediate codecs such as ProRes and DNxHR use intra-frame compression and much higher bitrates. They are designed for editing performance, whereas H.265 is optimized for acquisition and final delivery.
Using H.265 in production and delivery
H.265 is used at both ends of the video chain: many cameras and drones record in 10-bit H.265 to save storage, and streaming services, broadcasters and disc formats such as Ultra HD Blu-ray deliver it to viewers. It is commonly carried in MP4, MOV, MKV and MPEG-TS containers.
| Application | Resolution | Typical bitrate |
|---|---|---|
| Video conferencing and mobile | 720p to 1080p | 1 to 4 Mbps |
| 1080p streaming | 1920 × 1080 | 3 to 8 Mbps |
| 4K HDR streaming | 3840 × 2160 | 12 to 25 Mbps |
| Camera recording | UHD to 6K | 50 to 400 Mbps |
Editing long-GOP H.265 directly can be slow on systems without hardware decoding. Editors often transcode to an intermediate codec or work with proxies, then export the finished master to H.265 for delivery.
Limitations and common pitfalls
The weaknesses of H.265 are complexity, licensing and compatibility, rather than image quality. The following points summarize what practitioners should check.
- Encoding is slower than H.264 at equivalent presets, so software encoding of high resolutions needs substantial processing power.
- Licensing through several patent pools created uncertainty and limited browser support for years.
- Playback compatibility varies with hardware decoders, especially for 10-bit 4:2:2 camera files.
- Low bitrates cause smoothing: H.265 tends to blur fine texture rather than produce blocky artifacts.
- Long GOP structures make frame-accurate editing and scrubbing heavier on the computer.
In plain words
Raw video is enormous, so it must be packed tightly before it can be stored or streamed. H.265 is a smarter packing method than the older H.264. It notices which parts of the picture repeat or barely change and stores them very efficiently.
The result is that a 4K movie can look just as good while using noticeably less data. The downside is that devices need more computing power to pack and unpack it.
A real-world example
A small streaming service prepares a nature documentary in 4K HDR. Its earlier H.264 encodes required around 35 Mbps for clean 4K images, which many home connections could not sustain. The engineers encode the new master in H.265 Main 10 at about 16 Mbps for the top 4K rung, with lower rungs down to 1080p at 5 Mbps.
Viewers on smart TVs with hardware HEVC decoding receive sharp HDR images with far less buffering, while older devices fall back to an H.264 stream. The service reduces its delivery traffic for the 4K tier substantially while keeping visual quality equivalent in side-by-side review.