The Alliance of Open Media – 10 Months in

09/06/2016

How time flies.

201606-progress

About 10 months ago, the announcement of the creation of a new alliance caught me off guard.

Somehow, Google, Microsoft and a few other companies put their differences aside and decided to create the Alliance of Open Media. The intent – create royalty free video codec to rival H.265/HEVC. I’ve written about the Aliance of Open Media. It is time to revisit the topic.

A few things happened these last few months that are worth mentioning:

  1. We’ve learned more about the alliance – Jan Ozer  wrote a good progress report
  2. AMD, ARM and Nvidia joined the alliance
  3. Ittiam joined the alliance
  4. Vidyo joined the alliance

I am told work is being done on the actual codec itself. From the report Jan Ozer wrote, the following is apparent:

  • Baseline for the codec is VP10 (Google)
  • Most contributions of technologies on top of it come from Mozilla and Cisco; though I assume Microsoft is contributing there as well
  • Hardware vendors are putting their weight to make sure the algorithms used are easy to place in a hardware design
  • There’s a focus on GPU acceleration, which is important
  • Intent is to have it integrated into a browser by the beginning of 2017 and have hardware acceleration a year later

All the right moves.

ARM and Nvidia

Adding ARM and Nvidia is quite a catch.

ARM is in charge of the architecture of most smartphones on the market today, along with many of the IOT devices out there. Having them on board means that considerations for mobile and low power devices are taken into consideration by the alliance – but also that the work of the alliance will find its way into future designs of ARM.

Nvidia is where you find GPU processing power. They complement the attendance of Intel, brining the important GPU players to the table. In a recent whitepaper I’ve written for Surf, I touched the GPU issue briefly. I’ve done some research in that domain, and it does seem like the GPU is the best candidate to handle our future video coding – having GPUs relevant to this next generation codec fron the start is an important catch for the alliance.

Ittiam

Ittiam is a recent addition to the alliance.

I’ve had the chance to know Ittiam a decade ago, while competing head to head with their VoIP software. They have expertise in the multimedia space and in video compression, but they still are the smallest (or least relevant) player in this alliance. Having them is required to fill in the ranks and grow in numbers.

It would be nice to see others join such as Imagination Technologies (who are larger and a lot more meaningful).

Vidyo

Vidyo just join the alliance. On one hand, it surprised me. On the other hand, it should have.

Vidyo is collaborating with Google for a long time now in VPx and WebRTC. Recently it reiterated that with the work it is doing on VP9 SVC for WebRTC (you can find out more about it on a guest post Alex Eleftheriadis shared here on scalability and VP9).

Their addition to the alliance means several things:

  • Vidyo is making itself an integral part of every initiative related to future video codecs. This is a smart move, as it maintains its lead in the backend side and the smarts that is placed on top of SVC capabilities
  • This future codec will have SVC support in it, hopefully from the moment it is released to market
  • While a smaller company compared to the other members, the contribution of Vidyo to the alliance can be larger than many others of its members

Qualcomm

Qualcomm is missing.

So is Samsung.

And a few other smaller mobile chipset vendors.

I think it is their loss, as well as a missed opportunity.

They both should have joined the alliance at its inception.

Apple

Apple being Apple, they aren’t a part of it. Putting ads in the App Store and changing subscription revenue sharing models were more important to them, which is understandable.

The thing I don’t understand here is that Apple has removed most of its support in H.265. What does it have to lose by joining the alliance?

There are three paths available to Apple:

  1. Go with H.265. The current reduction in its support of H.265 can only be explained as a negotiation tactic in such a case
  2. Go with the Alliance of Open Media. Which it could do at any point in time. But if that is the case, then why wait?
  3. Release its own unique iCodec. Apple knows best, and it is time to lock its customers a bit further anyways

I wonder which route they are taking here.

Content Creators and Service Providers

We’ve got YouTub, Netflix and Amazon already covered. The internet may rejoice.

But what about Game of Thrones? Or the next movie blockbuster? Are they staying on the route of H.265 or will they veer away from it towards the alliance?

Hard to tell, though for the life of me, I can’t understand a long term decision of staying with H.265.

It would be nice to see the large studios and even Bollywood join the alliance – or at the very least back it publicly.

Timeline

If we look at the VP9 timeline, we havethe following estimates:

  • 1 year – Chrome decoding, along with a small percentage of YouTube videos supported
  • 2 years – First chipsets and reference designs support. My bet is on Nvidia and Intel here
  • 2.5 years – Chrome official support of it for WebRTC

H.264 in WebRTC

H.264 is hear to stay. More worrying – H.264 will grow in popularity in WebRTC services during 2016.

This progress and success of the alliance changes nothing in the current ecosystem and the current video technology.

The future of H.265

The future of H.265 does look grim. I do hope the alliance will kill it.

H.265 is in a collision course with VP9. It is still the more “popular” choice in legacy businesses, but that may change, as commercial deployments of it are small or non-existent.

The alliance simply means that a future codec is based on the VPx line of codecs instead of the H.26x ones. Now developers shifting from H.264 to a better codec will need to decide if they switch codec lines now or just later.

The royalty issues around H.265 along with the progress made in the alliance should tip the scales towards VP9 on this one.

What’s next?

Money time.

Where does that leave us all?

  • Vendors who handle codecs directly should join the alliance. The benefits outweigh the risks.
  • Consumers and users can continue not caring
  • Developers, especially those of backend media servers, need to decide if they shift towards VP9 or wait for the next generation to switch to a royalty free codecs. They also need to decide if they want to use VP8 or H.264 today

So… Until AV1 is here – which video codecs should you use in your application? Here’s a free mini video course to help you decide.