Menu
Home
Log in / Register
 
Home arrow Economics arrow Content Delivery Networks. Fundamentals, Design, and Evolution
Source

Compression

Much is talked about CoDecs. Even in this book I have visited the topic several times. There is a legacy within the sector, where the CoDec choice defined if you could reach an audience at all, to strongly prioritize CoDec choices as central to a content delivery strategy.

Today, in my opinion, there are two video and two audio CoDecs that are important if you want to reach an audience. These are h.264 with HEVC as the emerging successor, and aac for audio (with video) and mp3 for pure music delivery.

I am pretty sure that will have put the cat among the pigeons, but I have become fairly thick skinned to the repercussions of stating my opinion. Most of the so-called proprietary video CoDecs on the market today are little more than variations or optimizations of h.264. Even Google's VP9 is not getting the traction that Google would like, as is evidenced by their recent addition of h.264 to WebRTC support in Chrome. The fact is that h.264 has been extremely successful in much the same way that mp3 was 15 years ago. The universality of support is critical to ensuring that an audience can consume the content. Regardless of any optimizations and special benefits that other pretenders to the crown offer, the fact is that nothing compares in priority to the ability to reach the person (or device) wanting the content. And h.264 and mp3 have that ubiquity.

AAC has had a strong ride on the back of its early incorporation in the HLS standard. Although it is not as widely supported as mp3 as a whole (in part because of legacy devices), it has near-ubiquitous support in the devices that premium content consumers use, so I would think of AAC as the next generation for forward-looking deployments.

In the same way I would think of HECV as that evolution in video CoDecs.

As with the comparison between AAC and mp3, HEVC brings better quality at lower bitrates than h.264. As 4 k and 8 k video emerge (which both support, albeit with some variance and limitations), that bitrate difference will increasingly impact where optimization is important. It does require support in devices, and for this reason HEVC is already in common deployment in production and contribution, but less so in consumer facing workflows. It is, however, coming fast thanks - in part - to ARM and Intel taking great strides to include it in their recent chipsets.

This tighter compression makes HEVC a good consideration for mezzanine storage too. Where large video libraries are stored or cached behind “transcode on-the-fly” architectures (which may transcode the HEVC to h.264 between the storage and the consumer for compatibility), the HEVC library will require less storage capacity than the same library in h.264. This makes its continued uptake inevitable.

So, for the foreseeable future, h.264 is a solid platform and HEVC is a good option to explore; likewise, for audio channels in video, AAC is perfect - ubiquitous where the h.264 and HEVC are - but for very long-term, backward- compatible audio compression, mp3 is still unbeatable, even if bit for bit AAC compression provides better compression ratios.

I am not going to go on further here about CoDecs, since there are a number of points earlier in the book that dig deeper, but I will say one thing about some of the pretenders to these crowns: always ask yourself if they give you the same reach as the major CoDecs. If they don't, then you must do some serious digging into your economics, including the cost of rolling out the end user device support, before you adopt new compression technology. Always work with a ubiquitous CoDec that is “good enough” in preference to one that offers the world to a tiny audience.

 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel