The most anticipated news from Netflix : the ability to download movies and complete series for viewing offline, without internet connection. These early days of tests have shown that this new option is especially surprising how well it works and the few limitations of.
The question is how is it possible that it works so well? The videos do not take up too much (high quality hour and a half around 1.2 GB of space) and the definition of the content is really remarkable, so here must be a trick, many will say. There is, certainly, and is called VP9 a compression codec that manages to give more for less.
An almost omnipresent format
This open and royalty-free video codec was developed by Google in 2012 and is one of the most widely used video compression systems on the internet. It is basic fact YouTube, which Google uses this and the Opus audio codec to unite both in the container video / audio WebM.
This codec competes directly with other like H.264 and HEVC / H.265, and although Google promotes its use and for example is supported natively in Chrome and Android, not so with the iPhone or Safari, and Apple It does not seem particularly interested in supporting this compression system.
Opera, Firefox and even Microsoft Edge support it, which is important since it is one of the codecs used in integrated video content in HTML5 standard.
To that support software accompanying the support compression and decompression hardware : more and more chips offer that special support for this format, and indeed in the Wikipedia we explain how most current hardware platforms provide native support for hardware decoding Contents encoded with VP9.
The thing changes with compression: If we want to encode videos with this format (which is what Netflix does), we will need CPUs and GPUs with that specific support. Currently we could make in GPUs AMD Polaris series (480/470/460 RX), Intel processors (such as Skylake i7-6700 or i7-7700 Kaby Lake) the MediaTek Helio X30, or Samsung Exynos 8890 Octa 8 Curiously, neither Qualcomm nor NVIDIA offer solutions compatible with the hardware coding of VP9 content.
You may also like to read another article on FreshLookApp: Gfycat Loops, so you can create and share GIFs with this new app
An almost magical compression
To overcome this incompatibility with solutions from Apple, Netflix have a second option compression for these downloaded videos: in addition to VP9 also compress these videos with H.264 codec, and in particular, with the H.264 / AVC High Profile, Also used in the Blu-ray format and which is significantly different from the H.264 / AVC Main used in its streaming broadcasts.
As explained by those responsible for the implementation of these systems in Netflix, H.264 / AVC High offers “tools that increase the efficiency of compression”, which also makes VP9 to a greater extent. The savings is obvious compression versus H.264 / AVC Main, and according to these experts, these systems offer the same quality video with a bit-rate between 15% and 19% lower.
For VP9, saving becomes 36%, a truly bestial figure. Especially if we consider that this saving is accumulating in those millions of downloads that Netflix offers with this new service, and the impact on data centers used and the bandwidth for those transfers is logically key to those savings. The demands during compression and decompression are greater, but the results are evident.
Beyond VP9: Gutting Content by Sequence
In fact in Netflix go beyond, and in addition to the use of these compression codecs analyze each video in an increasingly complete way. Already we talked last year about how this compression varies depending on the type of content (not the same movie a cartoon action, for example).
Now Netflix divide each content pieces of one to three minutes, and computers to analyze these clips apply a different and adapted to the visual complexity of each compression.
The goal: to make the maximum possible quality, but also the size of the download is minimal. The improvements are evident, and Netflix is planning to bring all of these advanced compression methods to streaming their broadcasts.