H.264: Beyond The Hype
- By Steven Titch
- Oct 27, 2008
OK, you’re sold. After a year of anticipation and hype, IP cameras incorporating the H.264 compression standard are finally available. You know there will be bandwidth and storage benefits -- music to the IT department’s ear. But what else do you need to know?
For starters, all H.264 cameras are not the same. Second, the cameras on the market now represent just the first generation models that are limited by available processing power. Early implementers may not see a lot of difference between H.264 images and MPEG-4, but they will be getting a jump on futureproofing their surveillance systems.
The marketing mantra about H.264 has been half the storage and bandwidth for the same frame rate as MPEG-4; or conversely, twice the frame rate at for the same bandwidth and storage. H.264 itself comes under the MPEG-4 standards umbrella and is also known as MPEG-4 Part 10.
In layman’s terms, video compression algorithms replace entire streams of video data with simple instructions to the effect of “the next 10,000 bits are exactly the same as the one bit just sent.” Or, “Fill in the next 100 x 100 portion of the frame with the same 100 x 100 portion sent in the last image.” Naturally, then, a critical feature of any compression algorithm is how it well in condenses images where there is lots of motion.
A major innovation of H.264 is its use of variable block-size motion compensation. That simply means there is more precise segmentation of an image where there is movement in a portion of a frame. H.264 can segment an image block as low as 4 x 4 pixels. Further, H.264 uses predictive motion compensation, where it can instruct a video controller to display moving images based on frames sent both before and after. The result is that H.264 cameras allow much more of the stagnant part of the frame to remain compressed while retaining the necessary detail and resolution on the elements of the image in motion.
The 2008 ASIS International Seminar and Exhibits, held in Atlanta in September, was something of a coming out party for H.264. Vendors such as Arecont Vision, Axis Communications, IndigoVision, March Networks, Pelco, Sanyo and Sony were touting H.264 cameras of various types. And although vendors were high on the standard, many advised users to temper expectations, especially with the first generation.
“Most demonstrations compare H.264 cameras next to MJPEG, not MPEG-4,” says Ed Thompson, chief technology officer for DVTel, Ridgefield Park, N.J., which supplies video management software supporting H.264. “You’ll be lucky to see any difference between [MPEG-4] Part 2 and Part 10.”
The reason is that processing power has not caught up with the standard’s capabilities, Thompson adds. Texas Instruments’ newest DaVinci digital signal processor (DSP) for video will have a bandwidth accelerator, Thompson says, but will not be shipping until the beginning of 2009. That means current H.264 cameras still can’t take advantage of the complete toolset H.264 offers.
These tools include speedier motion searches. H.264, says Alex Swanson, program manager at IndigoVision, Edinburgh, U.K., is engineered to handle special types of motion particular to security, such as the image movement created by pan, tilt and zoom operation. IndigoVision’s cameras use H.264 to optimize PTZ search more effectively, Swanson says. But these tools add to the cost of processing. “There’s a layer of detail below H.264 that’s not really discussed,” he says. “The algorithm is clear. How you do it is up to the vendor.”
Other tools and features H.264 supports is a set of up to 15 analytics algorithms, including trip wire, unattended bags and crowd counting, says Peter Wilenius, vice president, investor relations and corporate development at March Networks, Ottawa, Ontario. H.264 will compete with other standards, Wilenius says, although it is particularly suitable to megapixel cameras because of the sizable amounts of video data they needs to process and transmit.
“H.264 is a necessary evil in megapixel cameras,” quips Tom Carnevale, president of Sentry360 Security Inc., Naperville, Ill., a megapixel camera supplier not yet supporting the standard. Currently, however, when images are placed side-by-side, Carnevale maintains there is no significant difference between H.264, MPEG-4 or MJPEG. H.264, he concedes, is a futureproofing mechanism. On the other hand, MJPEG, which Sentry360 cameras do support, “will remain a very good open architecture.”
On the server side, H.264 “delivering like it’s supposed to,” says Roger Shuman, marketing manager for Exacq Technologies, Indianapolis, which supports Arecont Vision’s H.264 camera suite. But like others who are high on the technology, he adds that there needs to be more processing power on the client side before users see a full payoff.
In addition, users should avoid viewing H.264 as if it were a magic bullet for all bandwidth, functionality and cost issues. “H.264 not completely understood by the market. There is not a wide selection of cameras, recording or management systems. It’s still in its infancy,” cautions Paul Bodell vice president of sales and marketing IQinVision. “There’s great bandwidth savings but it’s a processor hog.”
Most users, Sentry360’s Carnevale says, will likely adopt a mix of cameras after considering quality and cost trade-offs. The improved motion compensation aspects make H.264 cameras especially suitable for dense, high-traffic areas where there is a lot of activity against irregular backgrounds. An H.264 camera, on the other hand, would not add much if it’s going to be placed on a stationary mount to monitor a seldom-used back door.