Digital video compression became a mainstream consumer technology about 10 years ago when the emergence of digital satellite TV, digital cable TV, and shortly thereafter, DVD invaded homes. Since then, the popularization of digital high definition TV has boosted the trend further, delivering crisp pictures on large-screen televisions. All these technologies are based on the MPEG-2 standard, which, in 1993, crystallized from twenty years of research, into a solution that was ready for implementation.
More than ten years have passed since the publication of the MPEG-2 standard and Moore’s Law has continued to push the boundaries of semiconductor technology. It is now conceivable to build substantially more complex algorithms than MPEG-2, and achieve a low power solution that can fit high definition video into a palm-sized camera or camcorder. And all of this is accomplished with compression efficiency that can store HD content onto a minuscule solid state device.
Compression Solves a Problem: Bandwidth and Storage
A few years ago, facing the rapid advances in fiber optics and disk drive technology, many analysts thought that both bandwidth and storage would become free. So, you don't really need compression, right? Actually, the reverse turned out to be true. As progress in broadband networking and mass storage technology increased, the demand for high-speed content grew faster than was ever anticipated. At the same time, we see bandwidth getting increasingly scarce and the demand for ever-smaller portable storage getting increasingly stronger. This has resulted in a big push for bandwidth-efficient, compression-efficient technology. As an example, the supply of satellite transmission to the homes in North America is limited by the number of geostationary slots in the sky, but the demand for content continually increases, spurring the transition to high definition, local channel and video-on-demand programming.
Tape is Dead
The dominant storage medium for camcorders, to this day, has been the DV tape. At a rate of 25 Mbps, which results in 13GB per hour, this storage medium is quite inexpensive. Unfortunately, DV tape suffers from a number of significant drawbacks, both physically and logically. Physically, a tape drive is a fragile electromechanical system that is sensitive to environmental changes like temperature and moisture. In plain English, tape jams, and it usually jams at the wrong time. In addition, a much more fundamental issue is that tape is not a good digital medium because it lacks random access and search capability. Rewinding a tape is very tedious and introduces delays that are not compatible with indexing, linking and searching. Finally, 13GB per hour tape is an inefficient way of archiving video in digital form, and putting the tape in a drawer is the ultimate in inefficient archiving.
These are the reasons that, in the Japan market in 2005 as an example, the majority of the camcorder market has migrated to DVD, HDD or flash memory-based recording. The majority of the tapeless camcorders use MPEG-2 at 8 Mbps. The only advanced product currently on the Japanese camcorder market using tape is an HD camcorder, which records MPEG-2 content at 25 Mbps on DV tape. What the industry needs is a new compression technology that provides a significant compression gain over MPEG-2 and enables solid state recording of HD video. Fortunately, this new technology exists and is called H.264/AVC.
H.264/AVC
The H.264/AVC standard is a joint activity between the ITU (hence H.264) and ISO/MPEG (hence MPEG-4 AVC: Advanced Video Compression). It is also known as H.264, AVC, as well as "MPEG-4 part 10." H.264/AVC provides a compression gain over MPEG-2 of 2x to 3x at the most common operating points, and thus, has become the most anticipated video compression technology.
Ambarella’s No Compromise on H.264 Main/High Profile
MPEG-2-based high definition video is compressed on HDV tape at the rate of 25 Mbps. For the hybrid digital camera/camcorder device, the compression challenge is how to store high quality high definition video on flash memory-and archive the content on a PC or a hard drive. Nothing but a compression gain of 2.5x over MPEG-2 allows for the recording of one hour of high-quality digital HD on a 4GB flash memory.
Compression Quality
In order to meet this challenge, Ambarella had to depart from the “MPEG-2-like” approach to H.264 compression where the compression strategy is essentially MPEG-2 with an H.264 syntax. Instead, Ambarella has thoroughly implemented all the features that make H.264 a more powerful compression technique.
These features include:
The codec supports both interlaced 1080i format, as well as progressive 720p. It also supports constant bit-rate (CBR) and variable bit rate (VBR), allowing optimal use of solid state storage. The result is a single-chip HD codec that does not compromise image quality for low power consumption.
Low Power
While state-of-the-art video compression is necessary for a solid state camcorder, compression quality alone is not sufficient because a portable device will always be constrained by battery size and recording time. Ambarella has met the challenge of reducing power consumption to less than a Watt, a mandatory requirement for HD encoding, with its breakthrough compression technology.
Conclusion
With the development of the H.264 standard, it is now possible to achieve compression gains that make HD recording on solid state storage a reality. However, many factors need to be addressed when developing a technology that supports the requirements of a hybrid digital camera. Ambarella has factored in the trade-offs between high quality video compression and low power requirements of a portable device to develop the first HD H.264 SoC that compromises neither.
