Unraveling Video Compression: The Benefits and Drawbacks of H.265
By Tim Scally, SDM Associate Editor
H.265 is the latest iteration of compression standards that are based on reducing temporal redundancies in video feeds, or areas of similarity between frames. It has a number of enhancements as compared with its predecessor, H.264, that help it further reduce the bit rate of stream, and although it represents the market direction of video compression formats, there are some drawbacks to adopting this technology now.
Though the technology is complex and advanced, the concept is simple: If a camera is in a fixed position in a room, the background doesn’t often change, but objects walking through the room present changes between frames of video feed. So rather than continuously retransmitting the areas that don’t change, these types of compressions will just transmit the changes, overlaying them on top of the reference image. The act of omitting areas of similarity saves a tremendous amount of bandwidth compared with the compression video format that doesn’t reduce those temporal redundancies.
THE WAY OF THE FUTURE
H.265 began to be adopted in about 2015 and started to gain a good deal of attention in early to middle 2016, according to Brian Carle, director of product strategy, Salient Systems, Austin, Texas. “It is recognized as the direction of compression technology in the video surveillance industry,” Carle says. “When IP cameras were introduced to the market, the compression widely used was motion JPEG because it was a relatively simple process, and the CPU and the chipsets on the cameras were not powerful by modern standards. When the CPU technology became more powerful, cameras began supporting MPEG4, which was a huge advance. Then H.264 became the next standard, and now we can see the market moving in the direction of H.265.”
Carle says the market’s shift in this direction is a result of a perception — a correct perception — that using H.265-enabled products will save dramatically on camera bitrate, which translates into reduced bandwidth consumption for sharing video, and reduced storage consumption for retaining the video.
He estimates, based on the history of other processing technologies in this industry, that H.265 will be the gold standard for at least the near future. “We’re probably looking at a five-to-seven-year lifespan before a newer technology starts to become relevant in the industry.”
So if H.265 is the way of the future, is a clear advancement in technology and will save bandwidth, this is the technology that you should be choosing, right?
Not so fast, says Carle. He identifies three potential drawbacks to adopting H.265 right now.
The first is the complexity of navigating licensing issues. “One of the drawbacks that a lot of people, especially consumers, may not be aware of, is that H.265 is a licensed technology,” Carle says. “It’s not just available for manufacturers to use for free. There are several organizations that contributed intellectual property that makes up H.265. The same is true of H.264, so it’s probably easiest to compare it to H.264.”
In the case of H.264, all patent holders are represented by a licensing authority that collects usage information from manufacturers that make products that use the technology, and it collects the licensing fees and distributes royalties to these organizations. The organization acting as the licensing authority for H.264 is called MPEG LA.
In the case of H.265, however, not all patent holders are represented by one licensing authority. Carle explains: “So for a product manufacturer to properly license H.265, it’s actually very confusing, and some may argue it’s not even fully formed yet — for instance, if a product has H.265 support, you may not be able to identify and properly pay royalties to the different patent holders.”
Right now there are three licensing authorities representing patent holders who contributed intellectual property to the development of H.265: MPEG LA, HEVC Advance, and VELOS, which formed earlier this year to represent some of the patent holders. “So products that came out prior to April of this year were not necessarily fully licensed for all the different patent holders that contributed to H.265 technology,” Carle says, adding there may be some patent holders who aren’t represented by any of those three licensing authorities.
“The future remains to be seen with respect to what will happen with H.265,” Carle says, explaining that the greatest point of risk is to the manufacturer, but adding, “there have been notable cases of patent holders coming after end users instead of product manufacturers. That is a litigation strategy that has been used in this industry.
“The argument could be made that no H.265 product at this point is in complete compliance with the licensing requirements of all patent holders because they [the patent holders] are not all necessarily represented.”
The second potential drawback is higher CPU utilization. Viewing video or processing the video in an NVR requires substantially more computational power using H.265 than it does with H.264. “If you have a workstation that would normally view 36 cameras on the screen at a time,” Carle explains, “if all those cameras were converted over to H.265 cameras, you’re going to have to upgrade the computer to a higher power workstation so it can decode the H.265 video, which is more CPU-intensive than H.264.”
But as the saying goes, necessity is the mother of invention, and a positive development in this area Carle sees is increasing integration of co-processing technologies designed to take the burden off the CPU for more processing-intensive technologies such as H.265.
“We’re starting to see some VMS systems integrate with co-processing technologies,” Carle says. “An example of that is GPU integration, most notably with NVIDIA video cards. Using the GPU to process some of the H.265 takes a load off the CPU of the system. Ultimately, that allows the processing of more video on a computer. That can be cost effective.”
Rather than having to use two NVRs to accommodate the heavier processing requirements of H.265, users can install a more sophisticated video card in the system, and if the VMS is integrated with the GPU on that video card, they can process substantially more cameras with a single server.
So while the higher processing requirements could initially be a drawback, the integrations with co-processing technologies demonstrate a way the market is moving to process H.265 effectively and at a reasonable cost.
The third drawback is a current limited selection of H.265 products. This drawback is mostly a result of the first two. As the market moves toward H.265 and more technologies and integrations become available, and of course as the uncertainties about licensing issues are ironed out, product selection will increase.
H.265 will be the next video compression technology the industry will embrace, but the compelling argument that everybody should adopt it is not quite there. For now, however, Carle offers an alternative solution capable of delivering many, if not all the benefits of H.265.
“Although the bandwidth savings of using H.265 are significant,” Carle says, “you can realize very similar, if not equivalent, benefits using advanced H.264 codecs. An example would be Hikvision’s H.264+, or the Axis Zipstream codec. Samsung has a codec — they all have an advanced H.264 codec these days.”
Carle says there are many more advanced H.264 products than there are H.265 products, and so while limited selection is still an issue for H.265, it is not an issue for advanced H.264. “You have more product selection, you don’t have to worry about licensing problems, and advanced H.264 products don’t require the same processing load that H.265 products require,” Carle says.
So while H.265 offers clear benefits and advancements over its predecessor, you should only make the switch knowing the potential drawbacks. And maybe, for now, an advanced H.264 solution might make the most sense for you.