AV1 (AOMedia Video 1) is an open, royalty-free video coding format developed by the Alliance for Open Media and finalized in 2018. Among video codecs, it is designed primarily for internet streaming and web delivery, offering markedly better compression than H.264 and VP9 and broadly competitive or better efficiency than H.265 at comparable quality.
AV1 was created as a response to the complex patent licensing of H.265. Its goal is to give browsers, streaming services and device makers a modern codec that can be implemented without per-unit royalties, while achieving lower bitrates for the same visual quality.
How AV1 compresses video
AV1 compresses video with the same basic hybrid block-based approach as earlier codecs: it predicts each block from neighboring pixels or from previous frames, transforms the remaining difference and encodes it with entropy coding. Its efficiency comes from many more prediction and partitioning options and from stronger in-loop filters.
Frames are divided into superblocks of up to 128 x 128 pixels, which can be split recursively into smaller blocks of many shapes. Prediction modes include directional intra prediction, chroma-from-luma prediction, compound inter prediction and warped motion for zooms and rotation.
The table summarizes the main tools and what each contributes.
| Coding tool | What it does | Benefit |
|---|---|---|
| Large superblocks | Blocks up to 128 x 128 pixels | Efficient coding of flat areas at high resolutions |
| Flexible partitioning | Square, rectangular and T-shaped splits | Better fit to object edges |
| Chroma from luma | Predicts color from brightness detail | Fewer bits for chroma |
| CDEF and loop restoration | Remove ringing and restore detail | Cleaner image at low bitrates |
| Film grain synthesis | Removes grain and regenerates it at playback | Preserves texture without spending bits |
Film grain synthesis deserves special mention. Grain is random and therefore very expensive to encode, so AV1 can denoise the source, transmit only parameters describing the grain, and let the decoder add statistically similar grain during playback.
AV1 profiles and technical parameters
AV1 defines three profiles that set the allowed bit depth and chroma subsampling. Most consumer streaming uses the Main profile with 8-bit or 10-bit 4:2:0, which covers both standard and HDR content.
| Profile | Bit depth | Chroma subsampling | Typical use |
|---|---|---|---|
| Main | 8 and 10 bit | 4:2:0, monochrome | Streaming, web, HDR delivery |
| High | 8 and 10 bit | Adds 4:4:4 | Screen content, graphics |
| Professional | 8, 10 and 12 bit | Adds 4:2:2 | Professional and archival uses |
AV1 supports resolutions up to 8K and beyond, high frame rates, and HDR transfer functions such as PQ and HLG. Encoder settings usually include a speed or preset parameter that trades encoding time against compression efficiency, along with a quality target such as a constant rate factor.
AV1 compared with other codecs
AV1 sits in the delivery-codec family together with H.264, H.265 and VP9, and differs fundamentally from intermediate codecs such as ProRes and DNxHR. Delivery codecs minimize bitrate with long GOP structures, while intermediate codecs favor fast editing with intra-frame compression.
| Codec | Licensing | Relative efficiency | Main domain |
|---|---|---|---|
| H.264 (AVC) | Patent pools, royalties | Baseline | Universal compatibility |
| VP9 | Royalty-free | Better than H.264 | Web video, browsers |
| H.265 (HEVC) | Several patent pools | Clearly better than H.264 | Broadcast, devices, Blu-ray |
| AV1 | Royalty-free | Similar to or better than H.265 | Streaming, web, real-time video |
| ProRes and DNxHR | Vendor licensed | Low compression by design | Editing and mastering |
Efficiency gains depend strongly on content, encoder implementation and settings. Comparisons that report a fixed percentage saving over another codec should be read with care, because a slow, tuned encoder of an older codec can beat a fast preset of a newer one.
Encoding, decoding and practical use
AV1’s main practical cost is encoding complexity. Software encoders such as libaom, SVT-AV1 and rav1e offer presets from very slow and efficient to fast and real-time capable; SVT-AV1 is widely used because it scales well across many processor cores.
Decoding is handled by efficient software decoders such as dav1d and, increasingly, by hardware decoding in recent graphics processors, televisions, streaming devices and smartphones. Without hardware support, high-resolution playback can use considerable processor power and battery.
| Situation | Recommendation | Reason |
|---|---|---|
| Large-scale streaming library | Encode popular titles in AV1 | Bandwidth savings repay encoding time |
| Web video for modern browsers | AV1 with H.264 fallback | Wide support with a safe fallback |
| Older devices in audience | Keep H.264 or H.265 | Missing hardware decoding |
| Editing source footage | Use an intermediate codec | Long GOP AV1 is slow to edit |
AV1 streams are usually carried in MP4, WebM or MKV containers, and the same coding tools form the basis of the AVIF image format.
Limitations and common pitfalls
- Slow presets are very slow; encoding a long film at the highest efficiency can take many times real time.
- Hardware support is uneven across older devices, so a fallback stream is still needed.
- Film grain synthesis must be tuned; poorly configured grain looks artificial or blotchy.
- Keyframe interval affects seeking; very long GOP settings make scrubbing and editing sluggish.
- Royalty-free does not mean free of all legal questions, although its licensing is far simpler than that of H.265.
The simple explanation
AV1 is a modern way of shrinking video files so they stream smoothly over the internet. It packs the same picture into fewer data, so videos load faster and use less mobile data.
It is also free to use for companies that build browsers, apps and devices. The trade-off is that creating AV1 files takes considerable computing power, and very old devices may not play them smoothly.
How it is used in practice
A video platform hosts educational courses watched largely on phones. It encodes each lecture at 1080p in H.264 and in AV1, using SVT-AV1 at a medium preset with 10-bit 4:2:0 and a two-second keyframe interval for smooth seeking.
Players on devices with hardware AV1 decoding receive the AV1 stream; older devices fall back to H.264. In internal quality tests, the AV1 version reaches similar visual quality at a noticeably lower bitrate, so viewers on slow connections see fewer quality drops.
The extra encoding time is spent once per lecture, while the bandwidth savings repeat with every view. That balance is the typical reason streaming services adopt AV1.