Tutorials Developer

Debargha Mukherjee: The Codec You Use Without Choosing It

By Infinite Creation · Profile

At a Glance

HEVC compressed well and arrived with an unresolved licensing position, which for a browser or an open-source project is disqualifying regardless of quality. VP9 and later AV1 exist because of that. Mukherjee is the first author on the VP9 overview and one of twenty-one on the AV1 paper, and both formats now carry an enormous share of streamed video.

  • Lived: active from the 1990s, Indian-American
  • Anchor year: 2013, VP9 presented at the Picture Coding Symposium
  • Strand: overlooked

Career

1

The problem as it stood

A codec that cannot be shipped without a patent licence cannot be a web default. HEVC's terms were unsettled for years, which left the open web without a modern format at exactly the point video traffic was overtaking everything else.

2

What he actually did

Mukherjee led Google's VP9 work, presented at the Picture Coding Symposium in 2013 with seven co-authors, and later contributed to AV1 through the Alliance for Open Media. Both formats were built to be royalty free by design rather than by hope.

3

How it worked

VP9 and AV1 use the same broad structure as their licensed contemporaries: flexible block partitioning, better prediction, a transform, entropy coding. The difficulty was achieving competitive results while routing around patented techniques, which constrains engineering choices in ways a clean-sheet design does not.

4

What it made possible

YouTube served VP9 at scale for years, and AV1 now carries a growing share of streaming. Their existence also changed the negotiating position around licensed codecs, which is a effect standards people care about more than viewers do.

5

Where the credit landed

The VP9 paper names eight authors and the AV1 paper names twenty-one. Neither format has a single inventor and neither is usually attributed to a person at all. They are attributed to companies, which is accurate and unusual.

Known For

  • Royalty-free was a design constraint from the start, not a licensing decision made afterwards.
  • Twenty-one authors on the AV1 paper is the honest picture of modern codec work.
  • The existence of a free alternative changes the terms offered for the licensed one.

Background and Detail

These formats exist because of a licensing problem rather than a technical one.

Engineering around patents

Building a competitive codec while avoiding patented techniques constrains the design in ways a clean-sheet effort does not face. Some efficient tools are simply unavailable, and the work becomes finding equivalents that reach similar results by another route. That constraint, not raw compression research, is the distinctive difficulty of royalty-free codec design.

Where This Fits

VP9 and AV1 compress video for delivery, the modern transmission link in the chain set out in A History of Broadcast Technology.

FAQ

Q: Why does royalty-free matter so much?
A: Because a browser or an open-source project cannot ship a format it cannot licence, regardless of how well it performs. Licensing decides availability, and availability decides adoption.

Q: Is AV1 better than HEVC?
A: Broadly comparable, with the advantage that it can be implemented without licensing negotiation. For a streaming service at scale, that difference outweighs small efficiency margins.

Vocabulary

VP9 An open, royalty-free video codec released by Google in 2013, competitive with HEVC and deployed at scale by YouTube. It was designed to avoid patented techniques so that browsers and open-source projects could ship it.

AV1 A royalty-free video codec published in 2018 by the Alliance for Open Media, a consortium of browser, streaming and chip companies. It succeeds VP9 and is designed to be implementable without licensing negotiation.

Alliance for Open Media An industry body formed to produce royalty-free media formats, whose members include the major browser vendors, streaming services and silicon manufacturers. Its formation is what made AV1 credible as a default.

Entropy coding The final compression stage, which codes symbols in proportion to how likely they are, so common patterns cost fewer bits. Every codec in this corpus ends with a version of Huffman's or an arithmetic equivalent.

Sources

Every record below was fetched and read before it was cited here. Checked 26 August 2026.

Translate this page

Machine translation provided by Google Translate, on Google’s servers. We do not check these translations and they will get technical terms wrong. The English page is the authoritative one. Following a link sends this page’s address to Google. Your browser may also offer to translate this page itself, which keeps the request on your device.