Dear Reader,
News of H.265 (or HEVC - High Efficiency Video Coding) have been flooding the multimedia market with numerous firms announcing solutions on both the encoding and decoding side. But the standard is yet to see wide spread adoption to the extent that new handheld devices are yet to ship with a HEVC decoder.
Why hasn't a technology like HEVC gone viral? Is the industry entrenched and invested in H.264 and MPEG-2 to such an extent that switching costs to HEVC are too high? who is most likely to benefit with a high quality of experience and a high compression ratio?
The abstract below attempts to provide a possible answer to the above question, with some fairly modest arguments and reasoning.
Feel free to comment in the comments section or drop me a note on your views.
Regards,
Jyothin
Introduction
We first start by examining the flow of media in a digital network from the time it is created/contributed till when it is consumed. We speculate on the quality and compression requirements at each stage of a media flow i.e. in the contribution stage, distribution stage and consumption stage.
A Media Flow: is defined as the path that a media stream takes from the creation stage to the consumption stage. Examples include; 1) a local file playback is contributed and consumed at the same local node, 2) A YouTube video is contributed from a storage server, distributed via the Internet, passes through a network of nodes, and consumed on a client device, 3) the same YouTube video was contributed by a consumer on, perhaps, a handheld device and uploaded via the Internet onto YouTube storage servers., etc.
A Node: is defined as any element that is either used in the creation, distribution or consumption of media while at the same time being agnostic of the media type and/or mode of transport.
For the purposes of this blog, please refer to the 'Other Definitions' section below for additional definitions.
Media Flows
Media can be created at numerous ends of a media flow. Nodes that were traditionally considered as media consumers have now, also turned into media creators. With the advent of numerous services provided by companies like YouTube, Hulu, Vimeo, Facebook, etc. to upload, store, stream and download content, it has become more than easy for consumers to turn into creators. This shift in content creation has also seen a shift in consumption patterns of content that has traditionally been considered of type Broadcast.
Moreover, new market dynamics like OTT services are forcing service providers to monetize these services, which have traditionally been fuelled by advertising revenues.
Figure 4 illustrates an end to end media flow diagram as agnostic to where the content is created or consumed. Different flows have different quality, compression and storage requirements.
Offline media content, which traditionally accounted for a large portion of the media transport mode, is slowly being replaced with online transport modes. However, offline transport modes do not pose significant real-time encoding or transcoding challenges when compared to online modes of transport. They however do pose real-time decoding challenges.
Creator nodes are those nodes that either capture or create content or have access to pre-created content to send to a distribution node or consumer node. Media creators include nodes such as; (A) DSNG, studios; (B) handheld mobile devices (smartphones, tablets, etc.), video phones, PCs/laptops, and other such similar nodes within an Enterprise network.
A distributor node typically helps in only the transport of media, with or without a change in the quality/compression ratio of the media itself. Media distributors include nodes such as DBS, Cable, DTT, streaming servers/gateways, network switches, network routers, local headends, set-top box, residential/home gateway, DSL CPE, Wi-Fi router/access point, base-stations, and DVR/DVD players, etc.
| Figure 1: A comprehensive media flow diagram |
A consumer node typically consumes media by either displaying/rendering it for viewing purposes or storing for later viewing. Media consumers include nodes such as; A: digital televisions, B: handheld mobile devices (smartphones, tablets, etc.), video phones, PCs/laptops, and other similar nodes within an Enterprise network.
Quality vs. Compression
A break down a media flow into different stages helps in examining video processing requirements at each stage by plotting a two dimensional chart; video quality requirements on one axis, and compression requirements on the other axis. The table below describes the nomenclature used in this blog.
Finally with this categorization we see which elements in a media flow are best suited for the adoption of HEVC 'first' which will then lead to adoption of HEVC in other stages, with an overall compelling advantage for using HEVC.
| Figure 2: Quality vs. Compression trade off for various nodes in a media flow. |
- Creators (A): Typically would need to capture content at the highest quality to have minimum loss of information and detail. However, the captured content needs to be transported to a studio or post production center as efficiently as possible. For example, a live sporting event that is captured using high end cameras is transported via satellite to the studio for post processing and distribution. Hence I've positioned creators of type A with the need for high quality and high compression.
- Creators (B): Content captured via handheld devices for upload to video servers will need high compression ratios in order to shorten the upload time as well as lower costs for the consumer. Hence a need for higher degrees of compression is felt here while at the same time not sacrificing quality greatly.
- Distributors: Being agnostic to the content format, such nodes will need to optimize bandwidth usage and hence need high compression ratios to pack more channels in the same available bandwidth.
- Consumers: For the best quality of experience will need high quality content for playback and viewing. However, cost sensitive consumers will also need to keep an eye on their bandwidth usage to keep costs under control and hence the need for optimal compression.
- Mass Storage: If the marginal cost of adding an extra Gigabyte of storage is higher than the marginal increase in content that needs to be stored then the need would be for higher compression. If on the other hand the marginal cost of adding an extra Gigabyte of storage is lower than the marginal increase in content then compression ratios can be sacrificed for storing high quality content.
- Offline Distribution: Offline storage media being cheap it would make sense to storage high quality content for later viewing by consumers that anyway expect a high quality of experience from offline storage medium.
Conclusion
Maybe HEVC just needs that one viral video for it to go viral !!
Other Definitions
- Media refers to audio, video and graphics but in this blog the primary emphasis is on video.
- Media can be contributed/created at the any end of a media flow.
- Media can be distributed via numerous transmission modes but here we primarily focus on digital mediums and digital modes of transmission.
- Media can be consumed at any end of a media flow.
- An "Enterprise Network" is any closed network that can distinctly be isolated from the Internet.
- A "Short form" video is user-generated video and other video clips generally less than 7 minutes in length.
- A "Long form" video is content generally greater than 7 minutes in length.
- Video calling is video messages/calling delivered on fixed Internet initiated by smartphones, non-smartphones, and tablets.
- Internet video to TV is video delivered through the Internet to a TV screen, by way of an Internet-enabled set-top box (for example, Roku) or equivalent device (for example, Microsoft Xbox 360), Internet-enabled TV, or PC-to-TV connection.
- Live Internet TV is peer-to-peer TV (excluding P2P video downloads) and live television streaming over the Internet.
- Internet PVR is recording live TV content for later viewing.
- Ambient video are nannycams, petcams, home security cams, and other persistent video streams.
- Mobile video is all video that travels over a 2G, 3G, or 4G network.
References
According to Frost & Sullivan, the market for video transcoding will grow to $630 million by 2017, up from $264 million in fiscal year 2012.