January 05, 2015 by Romain Bouqueau

Passage from 2014 to 2015

Dear readers,

Happy new year! May 2015 bring you happiness and success :)

We just wanted to share with you our views on what happens in 2014 and what will happen in 2015. There are many retrospectives and futurology articles out-there. As Alan Kay said:

The best way to predict the future is to invent it.

So here is what we did and what we are going to do.

Retrospective on 2014

2014 has been a tremendous year for us. We made great demos (HPA retreat, NAB, IBC, ...) with big actors of the ecosystem involved (Elemental, Fox Labs, Akamai, EBU, AMD, ...). The company grew. It is a proof that free software can be business-friendly.

Our open-source project GPAC has modernized a lot. I think we finally started to take into consideration some important factors such as security. Modern development good practices improved. We also started our move to github.

We also kept on our efforts on the standardization side. It is really important to come up with clean standards. By standard, I also mean the reference software and test vectors. At a time where Test Driven Development (TDD) becomes the rule for software engineering, we should be careful not to keep on doing standards relying too much on a few clever - but human - individuals.

GPAC Licensing is a huge success. We try to innovate in our business model. I don't think it is stable yet, but in 2014 we have converged. Our business offer is now architectured around several units:

• Selling GPAC licenses: MP4Box, HEVC, DASH, Common-Encryption, the player for 4K and monitoring purposes are the main purposes.
• Selling licenses on custom software using some GPAC code. The Signals platform is the perfect example, more on this tool later this year :)
• Consulting on some multimedia open-source code. This goes really beyond the GPAC core (OpenHEVC, FFmpeg, GStreamer, etc.). There are 3 main activities:
  - Analyzing streams which don't work.
  - Fixing bug or adding a specific functionality.
  - Help to build a specific project (GPAC, VLC, etc.) with a specific configuration (tools, platforms, etc.). This led to our zenbuild open-source initiative.
• Trainings: we believe in education and we love explaining technologies.

Important themes for 2015

GPAC will keep on modernizing. We have new development such as a MP4 javascript demuxer called MP4Box.js, a build helper named zenbuild, and our C++11 platform called Signals.

For 2015, we wish to push more toward precision in wordings. We hope that people will cease to mix up technologies and their implementations (e.g. H264 and x264, or HLS and the Apple implementation).

I also want to work on innovative business models with our customers. With Signals, parts of our code are only open to our customers under a custom license. This way they can get the benefits of open-source (recompiling when the platform is obsolete, re-brand the product, etc.) and contribute with money (buying licenses) to the project. If you haven't watched it already, this talk from Marten Mickos is a must-see.

Here is an informal note we wrote a year ago for a customer. I think it is still quite relevant, although the streaming of subtitles (TTML/WebVTT ; read this and this) will also be an important subject:

Overview

The most important things are the most simple: the Web is changing our life. And it will bring most the upcoming technologies: protocols (WebRTC), HTTP streaming (HTML5/MSE), encryption (HTML5/EME), decoders (Google "Native Client" a.k.a. NaCl), subtitles (WebVTT).

The video industry used to be led by the broadcasters. New technologies will emerge using the Web. It doesn't mean that TV is dead. It doesn't mean that ISO will be replaced by W3C. It means that the broadcast industry has constraints while browsers are all software based.

Broadcasters and content providers meet the Web with OTT. One is still using MPEG2-TS, the others prefer ISOBMF/MP4 File Format.

New actors

New actors like Google will still make mistakes. But they will make less mistakes.

New actors like Google will make innovation that none of the existing actors can do.

New actors like Google, when they are big enough, can strike at any point of the workflow and invent new ones. They can buy companies or license technologies for everyone. They force a concentration of the other big existing actors.

[...]

Youtube will have features converging with television.

MPEG-DASH

There is a momentum for each technology. MPEG-DASH was awaited as the new messiah. But it was perceived mostly one more competing standard. It was promoted soon, but left a general feeling to have a standard which can even make coffee. After IBC 2013, some commentators said we were in a chicken and egg issue. 2014 might change things and here is why.

First there are the browsers; think of Youtube, or Netflix in IE11 using HTML5/MSE and HTML5/EME. Second there is a will to push DASH and the various initiatives are at a point where significant efforts such as players, compliance checkers, test sequences, content generators, etc. are getting more easily accessible.

The technological cost of MPEG-DASH is moving toward zero. The focus will be on the transport side. This is a killer for new companies trying to launch DASH services. But encoder manufacturers will still be in a position to charge their customers for the integration. We see that almost all our GPAC DASH user are companies. This can be an issue since it is a barrier to test easily DASH.

PS: as of 2015, we are part of DASH-IF and have brought their profiles to MP4Box.

Common Encryption and DRMs

Security is at the heart of the digital society. There is a risk that our society becomes locked or under a sort of global surveillance (more like Minority Report than 1984). [...] As said Voltaire: "With great power comes great responsibility".

DRMs in video are a good example of this duality. On one side it controls what distributors and final users are allowed to do (Rights Management). Sometimes in a restrictive way (a digital movie which must be watched on certain devices, not simultaneously, or simply reset after 48 hours, etc.). It forces States to set rules. This is particularly true in France.

DRMs also have a less visible side. Since the Personal Computer is also open, it means that we have to design closed paths. These paths exist in any ARM or Intel device. The Intel implementation is quite scary: an additional co-processor runs all the time (i.e. even when the computer is down) with higher privileges than the main CPU and runs all sorts of software (including a JVM). [...] I wish I can be wrong on this one. PS 2015: know more.

A less intrusive way could be to use watermarking. PS: as of 2015, MPEG has started the development of MPEG Variants.

Scalable codecs

SVC didn't succeed for several reasons: it arrived too late (i.e. when AVC decoding chips were already widely deployed), it implied an increased complexity, and the bandwidth improvements were not obvious. [...]

Its successor, SHVC, is already in a final draft stage (i.e. the specification shouldn't move until its publication). Its signalling is simpler. A fast open-source decoder (also very good for HEVC) called OpenHEVC is available. [...]

Scalable codecs allow to deliver a base layer to every viewer (e.g. HD video with stereo audio) and enhance the experience (e.g. UHD video and surround audio) without the need to resend the content. Only the improvement layer (e.g. the extra video pixels or the extra audio channels) needs to be sent. It opens new subscription models where users can choose what they like the most. PS: as of 2015, audio codec makers provide object-based codecs. This would allow to subscribe to a specific channel or mix songs the way you like. Another use-case of scalable somehow!

Hybrid delivery

With scalable codecs being available, hybrid delivery will develop (e.g. using broadcast and broadband like in HbbTV ; or mixing centralized servers (CDN) and P2P delivery (PS 2015: as Streamroot explain in their whitepaper)). We have many demos at GPAC. And we will push toward the standardization of some technology to help in this perspective. [...]

Hybrid is probably the only way to send HFR (high frame-rate), high bit-depth, high resolution (4K), HDR (compression of the dynamic range) to a wide number of people. [...] Given the acceptance of low quality in some household, we should expect a split between occasional TV viewers (who will never participate in social media interaction, nor buy a fancy new TV) and a geeky population who can afford to upgrade their equipment often. [...]

We will likely wait until the end of the decade to see a mature ecosystem on this.