3G Video Codecs
Mar 3, 2007 8:00 AM
As broadcasting moves to multichannel operations, maximizing the use of the available bandwidth becomes critical. Unfortunately, the current ATSC standards do not support mobile delivery. Because MPEG-2 needs relatively high bit rates for proper signal delivery, it is a less than optimum compression delivery for this application.
Although today’s MPEG-2 compression technology lowers the required bit rates for high-quality HD content, the bandwidth needed for broadcast quality still limits channel delivery options. Content that required 18Mb/s a few years ago now can be encoded with equal quality at around 12Mb/s. But this data rate is still too high for many emerging delivery channels, which must operate at much lower bit rates to fit within the allowed spectrum.
3G technologies
Third generation (3G) compression technologies are enabling content delivery over emerging distribution channels and allowing traditional channels to increase their services. Two newer codecs that are at least two times more bit-rate efficient as MPEG-2 have been standardized and are better suited for broadcaster use. These codecs include AVC (Advanced Video Coding) and VC-1 (Video Coder 1).
VC-1
VC-1 consists of three profiles: Simple, Main and Advanced. Simple and Main profiles were originally developed for use in networked computing environments using lower bit rates.
Because of this, certain assumptions were made regarding the display environment (e.g. square pixel aspect ratio) to improve compression efficiency. The Advanced profile adds in-band metadata support to optimize content presentation on a variety of display devices.
Advanced compression tools include:
- Multiple VLC tables that can be switched based on input video;
- DCT/IDCT transform switch, with multiple DCT block sizes — 4x4 to 8x8;
- Quantization (regular and dead zone);
- Fading compensation.
Both HD-DVD and Blu-ray have adopted VC-1 as a mandatory video standard. Windows Vista includes VC-1 decoding and related components needed for playback of VC-1-encoded HD-DVD movies. Additionally, VC-1 has been specified as the Xbox 360 video game console's official video codec.
Comparative compression performance ranges from two to three times that available from MPEG-2. Of course, this gain is dependent on the complexity of the content and available bandwidth.
| Video Format |
MPEG-2 | WMV9 Advanced Profile |
| 480p24 | 4-6Mb/s | 1.3-2Mb/s |
| 480i30 | 6-8Mb/s | 2-4Mb/s |
| 720p24 | 19Mb/s | 5-8Mb/s |
AVC
AVC/H.264/MPEG-4 Part 10 is built on MPEG-4 Part 2. The original three profiles were Baseline, Main and Extended. They have been supplemented by Fidelity Range Extension (FRExt) profiles.
The Baseline profile was designed for low latency, does not have B frames and omits CABAC (a lossless, probability-based data reduction technique). While the Main profile provides the highest compression, it does so at the cost of requiring significantly more processing. The result, however, is the highest quality video at the lowest bit rate.
New features in AVC/H.264 include:
- Intraframe prediction and coding — Neighboring macro blocks can be referenced;
- Interframe prediction and coding — Multiple block sizes can be examined for motion compensation;
- Variable vector block sizes — motion compensation can be done on blocks as small as 4x4 in conjunction with up to 32 motion vectors;
- Quarter pixel motion estimation;
- Adaptive loop deblocking;
- Integer transform;
- B frames can reference other B frames.
FRExt addresses the needs of professional HD-DVD production and lossless/uncompressed coding. It supports up to 12-bit color depth, 4:4:4 sampling and RGB color space. New features include adaptive residual block size and integer transform, 8x8 luma interframe prediction, quantization weighting and monochrome coding for black and white video.
Tests by the ITU Joint Video Team showed MPEG-4 delivered coding gains of one and a half times or greater for 78 percent of the 85 test cases. Fifty-one of the test cases showed coding gains of anywhere from two to four times.
For standards consideration
Both compression techniques are at various stages in the SMPTE and ATSC (www.atsc.org/standards/candidate_standards.html) standardization process. VC-1 has been standardized by SMPTE, while AVC references ISO/IEC 14496. Amendments to A/53 that define transport characteristics for both compression formats are in the review process.
The 493-page SMPTE 421M-2006, "VC-1 Compressed Video Bitstream Format and Decoding Process," defines the bit stream syntax, semantics and constraints for compressed video bit streams and describes the complete decoding process of VC-1.
Two VC-1-related Recommended Practices are available now. RP 227-2006: "VC-1 Bitstream Transport Encodings" describes the protocols for carrying SMPTE VC-1 video Simple, Main and Advanced profile elementary streams over MPEG-2 transport and program streams. RP 228-2006: VC-1 "Decoder and Bitstream Conformance" specifies the materials, procedures and criteria for verifying conformance of SMPTE VC-1 video decoders and bit streams. SMPTE sells licensed VC-1 test materials at www.smpte.org/smpte_store/test_materials.
ATSC Candidate Standards CS/TSG-658r1 and CS-TSG-659r2 define video system characteristics of VC-1 and AVC, and are under consideration as amendments to A/53. In these documents, the compression constraints for both AVC and VC-1 are defined as:
- 1080p60;
- All ATSC formats;
- 1440 x 1080;
- 544x 480, 528 x 480, 352 x 480 24p and 30i 4:3;
- 352 x 240, 176 x 120 24p 4:3.
An additional candidate standard, CS/TSG-660r1, proposes an amendment to A/53D to define transport of AVC compressed bit streams. A section is dedicated to the use of AVC in an E-VSB transmission implementation.
Implementation roadblocks
There are always issues with the introduction of any new compression codec, no matter how good the benefits. The first is backward compatibility. Unless a content supplier owns a closed-loop system (like that of DirecTV), changing compression schemes is virtually impossible. With other decoders/receivers in the field, you as a content supplier cannot typically dictate their replacement — unless you are willing to pay to replace them.
Second, any system in place has inertia. It is often difficult to get people to adopt new technology — no matter the benefits.
Fortunately, the ATSC system has the benefit of being modular, so new features can be added as needs develop. This attribute may allow the addition of new and improved codecs to support narrow band, portable and other new delivery techniques. We need not "throw the baby out with the bath."
Additional Reading
An excellent AVC/H.264 white paper from the SPIE Conference on Applications of Digital Image Processing XXVII: (www.cdt.luth.se/~peppar/kurs/smd151/spie04-h264OverviewPaper.pdf).
An overview of VC-1 from Microsoft: (http://www.microsoft.com/windows/windowsmedia/howto/articles/vc1techoverview.aspx).
A comparison of both codecs from an EETimes article:
(http://www.eetimes.com/news/latest/showArticle.jhtml?articleID=184417335).
| Want to use this article? Click here for options! |  |
