Cameras are increasingly becoming a bigger part of a human to machine interface technology, the way that machines interact with human and surrounding physical worlds to see , analyze and decide how and what to do next. They are also increasingly ubiquitous in cars, IoT and surveillance system, and game and entertainment devices seeing their resolution going as high as 4K and 8K.

Chips & Media Co. Ltd. is betting on the rapid proliferation of the vision-based machine interface and ultra-high resolution camera technologies for its second take-off, as it will open up new  market opportunities for video encoding IPs.

Transmitting pixel-rich vision data from cameras and image signal processors all the way through image and vision processing SoCs and then storing them on memory chips are increasingly becoming a tall order, because it squeezes memory space and data bus bandwidth. r memory spaces.  That’s where data encoding, or compressing technology works.

“When your device is flooded with visual data, it needs to encode them to transmit and store using less bandwidth and less memory spaces. In doing so, it eventually helps save power consumption,“ said Steve Kim, CEO with Chips & Media.

True enough, video traffics are exploding inside or between devices, squeezing memory space and bandwidth to the limit, as cameras are everywhere.

Especially fuelled by compelling, but high data rate killer apps like Big Data Analysis, in-vehicle ADAS, or advance driver assistant system, surveillance camera and IoT monitoring systems, ultra-high resolution cameras like full HD-and 4K-capable ones are now in wide use, gulping bandwidth and memory spaces.

That helps explain why video encoding IPs are in high demand.

Compressing, or encoding  video data is to identify and eliminate redundant image frames, or remove unnecessary or less important video information to save resources to store and transmit video data. Encoding IP, or circuitry blocks are usually embedded into multimedia SoCs to encode captures images block by block.

Key to determining the competitive edge in this video encoding technology is how efficiently to compress video in the smallest possible file size without compromising original or source code resolution quality. It is also important to use less of computational resources in doing that, as a stream of ever changing image elements are taking huge computational cycles to compress.

That’s where the forte of Chips & Media’s video encoding IP algorithm comes in.  “Our encoding algorithm is far better than competition at compressing pixel-heavy video data in the smallest possible file size not only keeping its resolution as high as that of source contents, but also saving resources in the  bandwidth and memory spaces of SoC ,” stressed CEO Steve Kim.

Pins high hopes on CFrame IP

Called as CFrame IP, for example, its video encoding IP portfolio is OS-and standard-agnostic, supporting multi OSes and multi video formats, and is so scalable and configurable that it can be easily customized to SoC vendors’ specifications.

It supports up to four color planes, a wide range of color formats such as RGB, YUV, Byer pattern, Monochrome, and 420/422/444 chroma sampling formats, for example. Its  compression rate is configurable to 1/2, 1/3, 1/4 times, or more by giving a target bit-depth to the original bit-depth, too.  It has a high performance capability processing 1pixel/cycle and it can be customized.

On top of that, the CFrame video encoding IP is a hybrid of lossy and loseless video compression technology, an integrated IP solution that can perform both of the lossless and lossy compression simultaneously just taking up the attributes of the two different compression technologies.

Underlying the hybrid structure is its own breakthrough 16×4 block-based   algorithm, which can tune the compression ratio of each block variably depending the image complexity. In doing so, it can take up juts the attributes of the two different compression technologies not only maximizing the compression efficiency, but also achieving  better image quality at the same compression rate than the previous IPs.

The intelligence enables it to compress 4K 30 or 60 frame source image data in the smallest file size available in the market without sacrificing image quality or resolution.

The company has already lined up lossless compression IP and lossy compression IP in each of its separate IP portfolio. As the resolution of source image data goes as high as 4K and 8K, however, toughest technology challenge facing SoC vendors is how to  reduce the bandwidth keeping the resolution loss at the minimum.

The lossy only compression IP couldn’t achieve good image quality, while lossless only compression IP couldn’t get constant bandwidth reduction. Yet, the CFrame IP works the both at the same time.  The CFrame IP the set of compression IP and decompression IP

Working on both – lossy and lossless compression

He sees huge market potentials in the rapid adoption of high resolution 4K security and surveillance cameras, drone cameras, 4K camcorders, and sports cameras, and ADAS, or advanced driver assistant systems.

Autonomous cars are a next wave of growth drivers that the company is zeroing in on for future growth, as they might come awash in a wealth of high resolution cameras to see around and make decision on which directions to go on their own.

Chips & Media is a sort of semiconductor IP licensor that licenses video codec, or encoder and decoder IPs for certain amounts of fees and is paid back royalties per chip if the IP-embedded SoC are shipped.

The video codec IP provider is little known outside the chip industry, but is often called as the Korean equivalent of ARM Holdings of the U.K. in the global video codec IP market, coming on top in the global video codec IP market.

Since its inception in 2013, its code IPs have been embedded into more than 440 million SoCs shipped as of the first half of 2016, addressing a wide array of IT markets from mobile phones to tablet PCs to digital TVs to STBs to digital still cameras to surveillance cameras to in-vehicle dashboards and ADAS. As the shipments of the SoCs, which are built with its IPs, are on the rise, so do its royalty revenues. In 2015 alone, 25 SoC vendors collectively paid 5.722 billion won in royalties combined. In the first half, the company chalked up 3.661 billion won in the royalty revenues, having licensed its IPs to 24 SoC vendors

Its list of licensees has been expanding to reach more than 80 companies, including such big name SoC makers or IT companies as NXP Semiconductors, Spreadtrum, Hisense, Samsung Electronics, Renesas, and LG Electronics, to just a few.

Looking forward, the company is scrambling to carve out growth momentums in the rapid proliferation of ultra-high bitrate and ultra-high definition video market, because that’s where there is fresh demand for new video codec standards like 4K H.265 and HEVC compression technology.

Resolution goes as high as 4K and 8K 

Especially, it pins high hopes on its s H.265 codec IPs, which was developed in 2013, as a growing list of source image data comes awash in several millions of pixels, or 4K and 8K resolution.

The rapid emergence of new video codec standards bodes well of for the future of the company. To save development and overhead costs of the new standards and cut short the time to market, SoC vendors are showing a growing trend of outsourcing video codec IPs to a IP licensor like Chips & Media, rather developing them on their own.

“We bet that H.265 video codec IPs would serve as a new wave of revenue growth driver at least over the next 10 years,” said CEO Steve Kim.

Yet, his ambition doesn’t stop there. He aims to remake Chips & Media as a total IP solution provider for video applications market that can supply ISP, or image signal processing IPs and image and video processing IPs. The line of new IPs would lay out the groundwork for its entry into autonomous car and drone markets.