Faroudja’s Bit Rate Reducer Gambit

There's no end in sight for the consumer's insatiable appetite for bandwidth. That demand continues to stress transmission networks and storage systems throughout the world.

But what if someone said there's a way to significantly reduce the bit rate requirements of video content -- other than current compression schemes -- without damaging the image quality?

That would make people sit up and listen -- especially when that someone is Yves Faroudja, a legend in the video industry. Faroudja has been behind almost every video processing and scaling technology development for decades and is a recipient of three Emmy technology awards.

So, of course, everyone in Las Vegas at this year's NAB (National Association of Broadcasters) Show listened when Faroudja Enterprises, a privately funded startup, trotted out its new technology, a video bit rate reducer designed to provide up to 50 percent reduction in video bit rates without reduction in image quality.

In a hotel suite at the NAB show in Las Vegas, Faroudja Enterprises showed off
its new technology, a video bit rate reducer.

The Faroudja's scheme doesn't alter current compression standards (MPEG-2, MPEG -4, HEVC). It's rooted in Faroudja's belief that such compression systems aren't using all the available redundancies to improve compression efficiency.

Under the new scheme, Faroudja introduces a new pre-processor (prior to compression) and post-processor (after compression decoding). "We take an image and simplify it; and that simplified image goes through the regular [standards-based] compression process," he explained. "But we never throw away information."

Instead, in parallel with the conventional compression path, Faroudja inserts what he calls a "support layer." This compresses signals not used in Faroudja's so-called simplified image. Together with the decompressed simplified image, the support layer helps reconstruct the original image in full resolution -- at a reduced bit rate.

System overview shows an example of how Faroudja Enterprises' process
works when applied to 4K video.

Faroudja claims "a bit rate reduction of 35% to 50% for an equivalent image quality [and] a significant improvement of the image quality on low bit rate content."

Details, of course, remain tightly wrapped in the black box. Faroudja hasn't decided on the initial market focus for his new technology. He said he's keeping his options open for a business model -- for now. Faroudja's team, however, has completed proof-of-concept and done a public demonstration.

Potential markets for the new technology include broadcast, cable and satellite, streaming media over the Internet, or Skype-like video conferencing.

Faroudja Enterprises is about to embark upon "market implementation" of the technology "over the next couple of months" by tailoring it to market demands, Faroujda told EE Times in an interview this week.

Jon Peddie, president of Jon Peddie Research, said that there is no one else who has rolled out anything similar to what Faroudja has done. "Faroudja is not displacing any company or technology. They are augmenting the existing codecs, including H.265," said Peddie.

But when it comes to its potential competitors, there are many. Richard Doherty, research director at Envisioneering Group, says future competitors are "universities with strong analytical math departments," including Cambridge, Oxford, Harvard, MIT, Columbia, Stanford, Princeton, Georgia Tech, U.C. Berkeley, Purdue, and the Fraunhofer institute.

Doherty predicts, "Once this is known to work, other low-key efforts might get ramped up." They'd be all racing to "receive funding to uncover other methods which do not infringe on Yves's.

Market Acceptance?
Faroudja's video bit rate reducer has many advantages. It's compatible with existing standards. The new process is compression-standard agnostic. It applies to all existing standards, from MPEG-2 to HEVC, and reduces bit rate in all cases, the company claims.

Faroudja explained that a final image will be preserved even if just the pre-processor is used, without the post-processor. The image survives even if the support layer is interrupted. Images with two levels of resolution can be made available at the same time -- for example, simultaneous availability of a program in both 1080 lines and 4K resolutions, according to the firm.

Moreover, Faroudja's support layer "can further be configured as a transcoder" to help convert video among existing compression formats, such as MPEG-2, H.264, etc., to and from Faroudja's support layer format "to save bandwidth, bit rate, or file size in the cloud without sacrificing image quality."

The Industry Reaction
After the NAB show, Faroudja reported that his company's technology demonstration in a Las Vegas hotel suite produced "a lot of very positive feedback" from audience members in the high-end video business, including broadcasters and cable and satellite operators.

However, still missing are "responses from streaming Internet companies and those with Skype-like teleconferencing technologies. We will know more [about the target market] after we attend the Streaming Media East conference in New York in May."

Asked about potential business and technology hurdles, Envisioneering's Doherty said, "If [the technology is] as simple as Yves suggests, not many. It would be a slam dunk for satellite, wireless, fiber, cable, anything." Asked who would embrace this first, Doherty predicted that bandwidth constrained back-hauls -- broadcast, cable, satellite, and cellular -- will benefit most.

Highly Dependent on Content
Tom McMahon at Del Rey Consultancy, a video expert, pointed out that the industry still needs to see how it works in a variety of scenarios -- fast motion video, noisy images, video in lower resolution, video streamed at a lower bit rate -- and how much gain it could achieve.

He noted that "a bunch of experts" discussed similar layered approaches a decade ago when the industry was still developing the H.264 standard. Dolby Laboratories worked on a similar scheme, he said, but the technology was never implemented. "The issue [of such an approach] is that it is highly dependent on content." However, he quickly added that service providers intent on bandwidth conservation and delivery of better quality video at lower bit rates are hungry for a technology such as Faroudja's. It's also applicable in smart encoder designs.

Satellite operators such as DirecTV, operating in a controlled environment, might be interested, said McMahon. Even next-generation set-tops or TVs/media boxes from Google or Apple could take advantage of meta-data offered in a support layer such as Faroudja's, he added.

In the end, market acceptance will depend on how much investment is needed for Faroudja's system (pre- and post-processing, and the inclusion of a support layer).

Jon Peddie pointed out that one of the initial hurdles in garnering market acceptance is "getting people to 'grok' it." He called it a "double-edged sword," because it is "black magic until Yves goes into detail." That's something Faroudja "doesn't want to do with anyone who isn't willing to commit to it. Anyone selling IP has to walk that balancing act: disclosure vs. trust/risk."

Peddie predicts that telecoms wanting to compete with cable and satellite should be early to embrace Faroudja's video bit rate reducer.

Faroudja explained that the processing power necessary for Faroudja Enterprises' process is roughly a "doubling of H.264." Most of the complexity, he noted, is in the pre-processor side.

In the technology demonstration, Faroudja used a multiple-unit rack system including off-the-shelf GPU boards. "We probably used 10 to 15 percent of the hardware [processing power] we had."

Asked if the technology is applicable to the consumer market -- including TVs, mobile phones, tablets -- he said it will be in the future, if it's licensed to chip vendors. However, he added that the process's present form isn't yet consumer-ready.

Faroudja hasn't exactly decided on the company's business model. Options he's considering include making the technology available in hardware, software, or through licensing. He said his company can do licensing agreements with system vendors, license IPs to silicon, or make its software downloadable on customers' hardware.

Faroudja Enterprises, which has developed its latest technology in complete secrecy over the last year, already has two patents granted and six more pending.

Faroudja's reputation as a veteran inventor over age 65 helped grease the skids for the patents. Faroudja was surprised that his patent applications were entitled to accelerated examination. Briefly assuming the aspect of an absent-minded professor, he said, "I didn't know that."

By Junko Yoshida, EE Times