A tight fit

Video compression is not new. Analogue TV has always used video compression to squeeze pictures of infinite resolution into finite fields and frames by means of discrete picture lines. However, the expanding demand for a wide variety of digital images to be delivered over network connections of differing bandwidths has amplified the need for a multitude of video compression devices, Steve Montgomery writes.

Colin Farquhar, CEO of Exterity philosophises: “Early, now legacy, systems using MJPEG and MPEG2 compression still work extremely well and we have massive volumes of MPEG2 decoders in set-top boxes and DVD players in the home, working away very happily. It takes a step-change in device deployment, such as the switch to HD transmission, to make fundamental changes to newer, more efficient or universal compression schemes, like MPEG4/H.264 with their greater scalability and suitability to a wide range of target devices.” What this infers, is that mass-appeal and market penetration is largely responsible for the success and longevity of a particular compression scheme à la VHS versus Beta.
“For surveillance applications, the ‘industry-standard' image compression format is JPEG - which is perhaps best known for digital still photographs” explains Yu Kitamura, Applications Solution Manger for Sony Europe “ In fact, using JPEG compression a network camera is acting rather like a digital camera taking 25 or 30 pictures per second. Often referred to as Motion JPEG or M-JPEG, it has relatively low processor demands and has enabled the current generation of network cameras. It is suited to monitoring applications where it is not always essential to provide a TV-quality frame-rate.
In the broadcast world, MPEG2 is now a legacy system for standard definition TV that is being replaced by H.264. H.264 is actually an advanced form of MPEG4, known as Part 10 AVC (Advanced Video Coding). It is becoming the de-facto standard for HD transmission, including satellite TV, hard disk recorders and Blu-ray players. However, convergence and divergence issues abound. Even within seemingly compatible standards there is incompatibility; MPEG-4 Part 2, used in DIVX, and Quicktime 6 is different to MPEG-4 Part 10. Even compression using a single standard such as H.264 has different levels, or profiles defining the level of compression and consequently size of the tool set a decoder is required to implement.
Fortunately the tendency of higher-end applications to converge onto H.264, is beneficial in that a single set of hardware and software tools will suffice for a number of applications. Colin Farquhar sees potential evolution of the video compression industry: “The split in applications between dedicated hardware single software applications eg STBs and general hardware multiple software applications eg PCs has, to a large extent, been detrimental to the growth of the industry, ideally we would prefer a marriage of the two. Hardware-oriented systems limit the device to a single format, whilst software-oriented ones have lower processing capability. Now we have the combined benefits of massive bandwidth and large amounts of processing power so that this can happen.”
A technique of this type is used by VBrick in their WM Reflector Appliances, as explained by Steve Kossar, Senior Product Manager: “Video portals can be set up manually or with integrated intelligence using our Reflector technology which will optimize network utilisation by taking a single unicast stream and rebroadcasting it locally via multicast or unicast at the right bit rate and compression format to users. It will serve the correct content from a stock of multiple versions without having to constantly transcode and use processing power.”
The video conferencing industry, is another market that is converging on H.264. “Along with other major videoconferencing equipment vendors, our codec of choice is H.264, but we have to maintain interoperability with other current and legacy systems throughout the world” explains Casey King, CTO of Lifesize, “During call set up there is a complex handshaking procedure that identifies the best application point to connect the two systems, which provides the most efficient and highest quality connection.”
Casey believes that that the trend toward H.264 is part of a long-cycle phenomenon: “Every ten years so we see a movement in this type of technology where market share changes radically. There are three driving forces that will cause this shift: much greater efficiency, which we have with H.264; error resilience, which is coming with high profile scalable video compression (SVC) and interoperability as more and more manufacturers deploy the standard across a wide range of devices, expanding the number of nodes that can be used for person-to-person communications.” Wayne Dunlap, CTO for the VSG group, Polycom agrees: “There are many methods of video compression. H.264 seems to have emerged as the clear winner, it is the closest thing to a universal coding technology for video. H.264 is so successful that there are hardware implementations of it in everything from camera chips, cell phone chips, and even some general purpose processors. It has such a lead and prevalence in the marketplace that it will be tough for other coding standards to overcome it. Scalable Video Coding allows one video stream to be encoded for high bandwidth and low bandwidth connections at the same time. H.265 will be the next-generation video codec.”
Digital cinema is an area in which picture quality is paramount, although here the constraints are upon longevity of content access and commercial factors rather than transmission or encoding efficiency which are reasons that this sector has not embraced H.264 in the same way. François Helt, Technology Application Manager at Doremi explains: “The seven major Hollywood studios formed the Digital Cinema Initiative (DCI) to agree on an encoding and display technology which was decided to be JPEG2000. This is free from licence royalties and is a visually lossless compression technology that will allow future archiving and re-encoding when a better standard emerges. The DCI set out to look further than the three or four years broadcast technology lifecycle, so films can be preserved for many years into the future; and we have to consider that a typical 35mm film will last 100 years or so.” A compression technology based on wavelets, rather than block processing of MPEG formats, JPEG2000 has different characteristics, François continues “It is flexible enough to be deployed at different levels and is considerably flexible and error resilient, so be used in high quality digital cinema without visible artefacts.”
It would appear that H.264 and its derivatives are becoming a de-facto standard for many video compression applications, although it is perhaps too early to declare it the outright winner. The next interesting area of consideration is how the H.264 encoded content is actually delivered. Steve Kossar: “New, highly innovative techniques are emerging to transport H.264 encoded video to end points, including RTP over the internet, Microsoft’s Smooth Streaming for Silverlight and Adobe’s Real Time Messaging Protocol (RTMP). These will radically affect the way that content is delivered and the extent to which new devices and applications evolve beyond the rapid rise that we have seen over the past three or four years.”
A further point, made by Wayne Dunlap, is that bandwidth is like CPU power, memory, and disk space. “No matter how much bandwidth you have, there will always be new applications that require more. If we look at the jump from the old CIF (352x288) to HD 720p (720x1280), there was a nine-fold increase in bandwidth requirements. Then HD jumped from 720 to 1080, which was another two-fold increase. Now people are talking about 2k video, which will be another four-fold increase. Bandwidth simply cannot grow fast enough.” So, it seems there will be a never-ending demand for new video compression technology to fulfill future applications.

Article Categories