Video Codecs Explained: H.264 vs H.265 vs AV1

What a codec actually is

A codec is a piece of software that compresses video. It takes a sequence of raw images — uncompressed, enormous, completely impractical to store — and produces a stream of compressed data that can be decoded back into something visually close to the original. The word itself is a contraction of *coder-decoder*: it does both jobs, encoding on the way in and decoding on the way out.

There are dozens of video codecs in active use, but for everyday consumer video, three matter: H.264, H.265, and AV1. They cover roughly the years 2003, 2013, and 2018 respectively. Each one was designed to do the same job as its predecessor with substantially fewer bits, and each one succeeded — at a cost.

This post walks through what makes each codec what it is, what the trade-offs actually feel like in practice, and which codec to pick for any given file.

H.264 / AVC: the universal default

H.264, also known as MPEG-4 AVC, was standardized in 2003 and has been the default codec for almost everything ever since. If you upload a video anywhere on the consumer internet, the platform's servers will re-encode it to H.264 at multiple resolutions, because H.264 is the one codec they can be sure every viewer's device will play.

The technical innovation in H.264, compared to its predecessor MPEG-2, was a much more flexible block structure (variable block sizes, intra-prediction within frames) and better motion compensation between frames. The practical result is that H.264 produces files roughly half the size of MPEG-2 at the same visual quality. That was a big enough improvement to justify a wholesale industry switch, and the switch was effectively complete by the early 2010s.

H.264's strengths today:

• Universal hardware decoding. Every phone, every TV, every laptop GPU, every game console made in the last fifteen years has dedicated silicon for decoding H.264. Playback is essentially free in CPU terms.
• Mature encoders. The reference open-source encoder, x264, is the result of more than twenty years of optimization. It is fast, predictable, and produces output that holds up against any competing implementation.
• Predictable file sizes. You know what you are getting with H.264. CRF 23 looks a certain way; CRF 18 looks another. There are no surprises.

H.264's weaknesses:

• Bit budget. Compared to newer codecs, H.264 needs more bits to hit the same visual quality. For 4K content especially, this is the kind of difference that adds up to gigabytes per movie.
• Patent encumbrance. H.264 is covered by a patent pool managed by MPEG LA. For most consumers this is invisible; for companies shipping products, it means licensing fees.

Use H.264 when you want maximum compatibility, when the file is going somewhere that will re-encode it anyway, or when you want to upload something to a platform that might be on flaky old hardware on the consumer end.

H.265 / HEVC: smaller files, complicated licensing

H.265, also known as HEVC, was standardized in 2013. It is the direct successor to H.264, designed by mostly the same people, and it succeeds at exactly what it set out to do: produce files about forty percent smaller than H.264 at the same visual quality, or equivalently better quality at the same bitrate.

The technical changes from H.264 to H.265 are incremental but substantial: even more flexible block partitioning (up to 64×64 instead of H.264's 16×16), more prediction modes, better entropy coding. The cumulative effect is meaningful compression gains, especially for high-resolution content where the larger blocks give the encoder more to work with.

The practical experience of using H.265:

• Hardware decoding is now universal on modern devices. Every iPhone from the iPhone 6 onward decodes H.265 in hardware. Most Android phones from 2016 onward do. Smart TVs from the last several years do. Browsers are increasingly catching up, though support is still uneven (Safari has it, Chrome and Firefox added it gradually).
• Hardware encoding is increasingly common. Recent Apple devices encode H.265 in hardware; recent Intel and AMD chips do as well. This matters because software H.265 encoding is slow — substantially slower than H.264 — and a hardware encoder is what makes H.265 practical for large workloads.
• Storage savings are real. A 4K movie that would be twenty gigabytes in H.264 is comfortably under twelve in H.265 at equivalent quality. For people who archive a lot of video, this changes the storage math.

The complication is patent licensing. H.265 is covered by three separate patent pools with different licensing terms, and the resulting confusion has made some platforms slow to adopt it. YouTube, for example, does not stream H.265 to most viewers. Web browsers were late. Open-source projects have been wary.

Use H.265 when storage size matters, when you are targeting modern hardware, and when you do not need to stream to the broadest possible audience. Use it for personal archives, for home media servers, for sharing to devices you know support it.

AV1: the open future, slowly arriving

AV1 was finalized in 2018 by the Alliance for Open Media, a consortium that includes Google, Mozilla, Apple, Amazon, Netflix, Microsoft, and most other companies that ship video to consumers. The motivation was explicit: build a codec that delivers H.265-class compression efficiency without any of the patent licensing complications.

AV1 achieves its goal. The compression efficiency is roughly twenty percent better than H.265 — a 4K stream that needed twelve gigabytes in H.265 might fit in nine and a half in AV1. The licensing is genuinely free for the foreseeable future, backed by the patent commitments of the consortium members.

What AV1 has, in 2026:

• Hardware decoding on most new devices. iPhones from the iPhone 15 onward decode AV1. Modern Android, modern smart TVs, recent Intel and AMD GPUs all have decoder support. The list of "things that play AV1 in hardware" has gotten long.
• Streaming use in production. YouTube serves AV1 to viewers who can decode it. Netflix streams AV1. The major platforms are using it where their users' devices support it.
• Hardware encoding becoming common. Intel Arc GPUs encode AV1 in hardware. Recent Nvidia (RTX 40 series and newer) and AMD GPUs do too. Without a hardware encoder, software AV1 encoding is extremely slow — much slower than software H.265 — but with one it is practical.

What AV1 lacks:

• Universal compatibility. Older devices simply cannot play AV1. The codec is genuinely new and the install base is still building out. Sending an AV1 file to someone on a 2018 phone will fail.
• A mature software encoder ecosystem. The reference encoders (libaom, SVT-AV1, rav1e) are improving rapidly but are not at the maturity of x264 or x265. Encoding speed and quality are still moving targets.

Use AV1 when you control the playback environment and storage cost matters more than upfront encoding time. Use it for streaming to platforms that prefer it. Avoid it when you need to ship a file to an unknown audience or to legacy hardware.

A side-by-side summary

ConcernH.264H.265AV1

------------ Year standardized200320132018 Compression efficiencyBaseline~40% better~50% better Hardware decodingUniversalModern devicesMost new devices Software encode speedFastSlowVery slow Hardware encodersEverywhereCommonBecoming common Patent royaltiesYes (MPEG LA)Yes (three pools)No Best forUniversal compatibilityModern archives, storageFuture-facing streaming

How to pick, in one paragraph

If you do not know which codec to use, use H.264. It is the right answer for any file going to social media, to messaging, to email, to embedding on a website, or to a recipient whose device you cannot verify. Move up to H.265 when storage matters and you control the playback hardware — your own archive, your media server, files you ship to known-modern devices. Move up to AV1 when you are streaming at scale, when the playback environment is controlled, or when you genuinely care about being patent-clean. For most people, "use H.264 by default, use H.265 when storage matters" covers ninety-five percent of cases.

Doing the conversion

All three codecs are available in the browser-based FFmpeg converter on this site through the advanced argument field. To pick a codec, pass the appropriate -c:v flag:

• H.264: -c:v libx264 -crf 23 -preset slow
• H.265: -c:v libx265 -crf 28 -preset slow
• AV1: -c:v libsvtav1 -crf 35 -preset 6

The CRF values are deliberately different because each codec uses a different scale. Roughly equivalent visual quality lands at CRF 23 for x264, CRF 28 for x265, and CRF 30–35 for SVT-AV1. Adjust to taste.

For more on what these flags actually do, see how FFmpeg works. For practical compression guidance, see Compress a video without losing quality.