Low-complexity compression solves video challenges

Barco Silex releases hardware implementation of world's smartest VC-2 low-compression standard

Nov. 12, 2015, Nov. 12, 2015 – 

Advances in image sensors and camera technology promise to make your video applications more sensitive, with better-quality images at higher frame rates. However, this also means that more bits have to be streamed, analyzed, kept in memory, or stored for archiving. In short, you'll need more bandwidth, expensive cabling, and new storage solutions. A smart solution to avoid these challenges is to use a lightweight, mezzanine compression, a compression that allows transporting and storing video at high quality but also at a reasonable cost, possibly even on your existing equipment. To help you do that, the video engineers at Barco Silex have recently implemented what is arguably the world's smartest low-compression standard - VC-2 LD (Low Delay)- into a lightweight, high-quality IP block for inclusion in your application's FPGAs or ASICs.

Serving high-quality video without additional infrastructure costs

Whether your applications serve the broadcast and professional audiovisual industry, or your domain is medical, transport, surveillance, military... you will be confronted with a tsunami of new video bits, and with the consequences and costs for your infrastructure. Take the local media distribution, for example, which is just moving to IP-based cabling and away from costly cabling and switches. Or consider industrial video applications that more and more depend on Ethernet-based networks, applications e.g. using one or more video sources for display, analysis or processing of images (conference rooms, surveillance, medical, video production, KVM applications, digital signage...). They are changing to network infrastructure that is cheaper, is standardized, readily interfaces with most equipment, and that is good enough for most video applications.

Good enough ... but not for the emerging high-quality formats such as HD or UHD video.

Most installed Ethernet networks today allow a speed of either 1 Gbit/s or 10 Gbit/s. But a 1 Gbit/s connection won't allow you to transport HD (high definition) video, a format fast becoming a standard. And the emerging UHD (ultra high definition) video at 60Hz requires up to 12 Gbit/s - too much for even a 10 Gbit/s Ethernet.

So should you then install a new, expensive network with even higher bandwidth, e.g. 40Gbit/s? Not necessarily. With VC-2 LD compression, it is possible to transport up to two 1080p30 videos in parallel over a 1Gbit/s link. With a 4:1 compression, you may transport up to 3 channels of 4k60 video in parallel on a 10 Gbit/s link.

Looking at the various codecs available for mezzanine compression, VC-2 LD turns out to be especially suited for today's video applications. It is a lightweight, low-cost solution that can be either added to the existing hardware or run in software. It has been accepted as standard (SMPTE 2042), which ensures equipment supplier support and thus future interoperability. And most important, it has been designed for low sub-millisecond delays, which is essential as many video applications include real-time components.

Figure 1 - VC-2 LD allows transporting high-quality video over prevailing IP networks

Other use cases: life events broadcasting, real-time surveillance, lossless archiving

A second use case where a light compression can make the day is broadcasting, especially of life events. With today's cameras, you may capture high-definition video formats. But the legacy cabling and transport, in which you have heavily invested, is not able to transport that video in highest quality into the production rooms, and on to the viewers. Full HD video, e.g., may only be transmitted over the newer and shorter 3 Gbit/s 3G-SDI links. But with VC-2 LD, the bit rate can be halved to make it suited for 1.5 Gbit/s HD-SDI links. These cables are cheaper and they allow longer cable runs, needed e.g. in outside broadcasting.

Figure 2 - VC-2 LD allow transport of HD video over your legacy SDI cables

Another use case is offered by dedicated cameras in industrial applications (e.g. food sorting, machine inspection, robotics, automotive...) or in surveillance and military equipment. These sometimes use sensors with a high frame rate and/or a high-resolution, capturing images for data-intensive processing and archiving. But chances are that you can only use a fraction of the potential, because the camera has to tune down its quality to fit into your video interface and storage capacity. With 4:1 compression in the camera, however, you'd be able to stream considerably better images over the same infrastructure, use a higher frame rate, or access the images much faster. And all that without any additional costly investments. That is, of course, if the 4:1 compression encoder is so lightweight that it fits onto the camera's existing FPGA. And if you can add a decoder onto your image processing board without having to redesign it, or, alternatively, add a real-time encoding software library to your video processing software.

Figure 3 - VC-2 LD lowers the bandwidth between your camera and viewing/processing

A last use case involves storage or archiving. Consider for example medical images or surveillance. Some of these images may have to withstand scrutiny in court, or will be used as proof, so the quality of the stored images has to be top. In some cases this even involves mathematically lossless storage of raw camera images. Again, the 2:1 to 4:1 image compression offered by VC-2 LD can save the day, allowing you to use lower-cost storage (disks, DDR external memory...) and still archive high-quality resolution pictures. Looking at video broadcasting, this means that you could retrieve the best viewing experience for any circumstance, with no concessions due to the available storage infrastructure.

VC-2 LD - a compression codec with some pretty unique qualities

VC-2 LD is a subset of VC-2, which was designed at the BBC's R&D center as a general-purpose compression codec. In contrast to VC-2 itself, VC-2 LD was specifically designed to enable low compression (2:1 to 4:1) with as little delay as possible, down to a few lines. And because of its origin at the BBC, it is one of the smartest codecs available; other comparable codecs may share some of its characteristics, but not all. VC-2 LD is a wavelet-based codec, using the same mathematical wizardry also used in JPEG 2000 and ensuring high-quality compression. But as VC-2 LD targets low compression rates, it avoids the high complexity of JPEG 2000. In addition, the VC-2 LD profile was designed as a line-based intra-frame codec, so that it achieves latencies down to a few image lines. Long latencies may be acceptable for archiving and end-user viewing, but they are unacceptable in many industrial or professional applications, let alone in live broadcasts.

Figure 4 - Comparing VC-2 with other popular codecs for compression and latency

The experts who designed VC-2 LD went to great lengths to make the codec as flexible as possible, so that it could serve a wide range of applications. Whether you target the best viewing experience or the highest information content, or VC-2 LD can deliver. It provides unrestricted image size (large and small), unrestricted bit depth, and 4:4:4, 4:2:2 and 4:2:0 color subsampling. And it allows setting number of parameters to find the best trade-off between compression efficiency, complexity and latency for any given application. Despite, the codec is straightforward to use: it provides, e.g., a number of default settings for common video formats, called the base video formats.

Last and possibly most important, a key distinguishing feature of VC-2 LD is that it is a standard and open technology, designed to avoid patent infringements. So it may be included in video production equipment without the potential costs, legal uncertainties, and practical difficulties of including other comparable codecs.

Now available as a lightweight IP block for the industrial market

Being a low-complexity codec, VC-2 LD can be implemented on your FPGA or ASIC designs without the need to upgrade to more expensive versions or larger footprints. But implementing a specialized video codec may require you to hire expert resources and to extend your project's timeline. A solution is to source the VC-2 LD compression as a ready-to-integrate IP block. This inroad wasn't available until recently, when Barco Silex' one of the world's premier specialists in video IP - announced it would add VC-2 LD to its library of high-quality IP blocks for the professional video market.

Barco Silex' engineers kept true to the design goals of VC-2 LD and created an IP block of minimal size and complexity, a block that can be integrated on your FPGA of choice, even if that is a low-range chip. The company ships its blocks as with a functional simulation test bench, synthesis and implementation scripts, and full documentation. And contrary to some other vendor's blocks, these have been made with easy integration in mind. In addition, the experts at Barco Silex are ready to assist you, be it to discuss your ideas, help draw up the most efficient design, even down to executing the full project.

Conclusion

High-quality video formats are coming! They offer better viewing, more information, and more possibilities. But implementing them may bring about a rather costly overhaul of your infrastructure. Mezzanine compression can save you money and effort. With the high-quality, low latency, low-footprint IP block from Barco Silex, adding VC-2 LD to your hardware toolbox has now become a cost- and time-efficient way to boost your video applications. With VC-2 LD compression thus included on your equipment, new options become available, without the need to invest heavily in new infrastructure and storage.

Developed at the BBC labs and successfully used at the Beijing and London Olympics

A decade-and-a-half ago, and seeing all the upcoming technical advances in high-quality video, the BBC considered existing compression options. They weren't impressed with what was available at the time: no codec seemed to match all their requirements. What they were looking for was a compression that had at the same time a very high quality and a low form factor and complexity. JPEG 2000 obviously met the quality requirement, but at the cost of too much overhead and complexity. The BBC was transmitting a lot of live coverage (and already eying the 2008 Beijing Olympics). So they needed a system that could compress video with only a few lines of delay.

As they couldn't source a system that fit their bill, the BBC experts set out to define their own compression system. The outcome of this R&D was a compression codec called Dirac, named after the English theoretical physicist and 1933 Nobel Prize winner Paul Dirac. With Dirac available, the BBC successfully delivered HD video of the 2008 Olympics, transporting it from HD-SDI to SD-SDI using a 6:1 compression ratio. The BBC being a public institution, they chose Dirac to be an open technology, free from royalties. And with their support, a version of Dirac called 'Dirac Pro' was soon standardized as SMPTE 2042, shorthand VC-2. The designers of VC-2 had the foresight to also include a low delay version (VC-2 LD), witch has an extra low latency down to a few lines and is especially suited for compressions of 2:1 to 4:1.

Barco Silex has its headquarters in Louvain-La-Neuve (Belgium), and offices in Ghent and Aix-en-Provence. Its 50 experts build IP blocks and offer FPGA and ASIC services, specializing in video and encryption (www.barco-silex.com). Barco Silex is a 100% daughter of Barco, the global leader in digital cinema and display technology.

Technical contact: Jean Marie Cloquet

Commercial contact: Geert Decorte:

www.barco-silex.com