Skip to main content

Acrimonious constraints

Acronyms are a form of compression. They are the staple of standards organizations, video manufacturers and industry trade publications like Broadcast Engineering. For example, we are currently in the middle of the DTV transition, from analog NTSC to digital ATSC.

Remember Table 3 in the ATSC A-53 DTV standard? The table lists the video format constraints on MPEG-2, the compression standard on which ATSC is based. (See Figure 1 below and “Web links”)

Vertical size value Horizontal size value Aspect ratio Frame rate Progressive sequence 1080 1920 Square samples, 16:9 display 23.976Hz, 24Hz, 29.97Hz, 30Hz Interlaced scan 29.97Hz, 30Hz Progressive scan 720 1280 Square samples, 16:9 display 23.976Hz, 24Hz, 29.97Hz, 30Hz, 59.94Hz, 60Hz Interlaced scan 480 704 4:3 display, 16:9 display 23.976Hz, 24Hz, 29.97Hz, 30Hz, 59.94Hz, 60Hz Interlaced scan 29.97Hz, 30Hz Progressive scan 640 Square samples, 4:3 display 23.976Hz, 24Hz, 29.97Hz, 30Hz, 59.94Hz, 60Hz Interlaced scan 29.97Hz, 30Hz Progressive scan Figure 1. The restraints on compression format from ATSC-53.

It is the only portion of the ATSC standard that the FCC did not include when they adopted the ATSC standard for DTV broadcasting in the United States. The standard, as adopted, constrains the use of MPEG-2, a compression standard that now seems almost as dated as the analog NTSC broadcast standard it's replacing.

The table includes 18 DTV video formats — technically 36, if you consider the six frame rate codes explained in the legend. These codes may be the ATSC's idea of data compression, but something tells me codes 2, 5 and 8 won't be used. And what happened to codes 3 and 6 — or MPEG-3 for that matter? Frankly, I haven't a clue, despite having been a participant in the ATSC meetings where the DTV formats were determined.

The previous century's legacy of interlace and noninteger frame rates lives on in this table, as do the ATSC format constraints that the FCC agreed were unnecessary. There was some talk about getting rid of the legacy NTSC frame rates at the end of the digital transition. Don't hold your breath.

MPEG-3 was supposed to be the standard for encoding HD video until support for HD was added to the MPEG-2 standard. Support for HD video was added to MPEG-4 as well — despite the original intent of MPEG-4 being a standard for low bit rate encoding, such as for videoconferencing, Web and mobile video.

MPEG-4 Part 2 defines the video compression algorithms for the standard. It has been amended twice, adding support for high-quality HD video encoding. The MPEG-4 Part 2 Studio Profile is used in the Sony HDCAM SR format, introduced in 2003, to capture the full 10-bit 4:4:4 RGB outputs from a CineAlta HD camera.

MPEG-4 Part 2 is also dated, compared with MPEG-4 Part 10 (also known as H.254 or AVC), which offers a 50 percent bit rate reduction for the same image quality. U.S. broadcasters can use AVC in a range of HD video acquisition systems. Unfortunately, they cannot use AVC to improve the quality or quantity of the video programs emitted from those DTV transmitters. The ATSC standard and a growing installed base of DTV receivers constrain broadcasters, even as competitors begin the transition from MPEG-2 to AVC.

The don't-convert program

The new advancements in HD video encoding can be used by broadcasters as they upgrade to HD production. The constraints of HD video encoding primarily apply to the bits emitted to those who do not subscribe to a multichannel TV service — the people who are still watching free-to-air TV.

In March, the NTIA released details of a government coupon program for the digital-to-analog boxes that will convert the DTV formats back into the interlaced 59.94Hz used by millions of legacy NTSC receivers. The legacy of interlace — noninteger frame rates and MPEG-2 — is being propped up by this converter program. To be fair, though, analog cable tiers are not likely to disappear by 2009, either. Meanwhile, the transition to digital HDTV continues to be as acrimonious as ever.

Consider the highly successful DTV transition in the United Kingdom. Almost half of the UK's 60 million TV sets are connected to a digital device. Most of those TVs are receiving Freeview, a free-to-air DTV service, or BSkyB, a DBS service and a Freeview partner, offering three of the DTV service's 40-plus channels.

In February, BSkyB announced intentions to use its bits on Freeview to deliver an upgraded pay-TV service. The service would use AVC compression to pack four channels in the space now occupied by three existing channels. The new programming would include movies, entertainment and live Premiership football. Subscribers would be required to purchase a new receiver with AVC and conditional access capabilities, in addition to the MPEG-2-based Freeview capabilities.

The other Freeview partners — BBC, ITV and Channel 4 — are reportedly set to tell media regulator Ofcom that BSkyB should no longer be a shareholder in the consortium. They say that with this move, BSkyB would no longer be supporting Freeview's commitment to providing UK consumers with access to as wide a range of free-to-view DTV channels as possible.

The BBC has conducted trials of HD DTV broadcasts using AVC compression. It is expected that a portion of the spectrum that will be returned after the shutdown of analog services between 2008 and 2012 will be used to support HD broadcasts.

Meanwhile, USDTV, a company offering a pay-TV service using digital capacity leased from U.S. broadcasters, is shutting down operations, forcing its 7000 subscribers to consider other multichannel offerings. USDTV followed the model of the original UK DTV service, OnDigital, a pay-TV service that was replaced by Freeview in 2001. This leads one to ask whether a Freeview-style service using AVC compression technology could attract viewers in the United States.

So long (GOP) HDV?

I believe two key issues will shorten HDV's lifespan: the limited quality using MPEG-2 MP@ML at 25Mb/s, and the complexity and quality implications of editing long-GOP MPEG-2. Despite these limitations, HDV has helped to bring HD acquisition to the masses, with a wide range of products costing less than $10,000. Now, it too is ready for replacement, thanks to AVC.

Last July, Sony and Panasonic set aside their sometimes acrimonious relationship to develop AVCHD, a consumer format that will use long-GOP AVC encoding to capture HD in both the 1080i and 720p formats. Many camera and editing system manufacturers have announced support for the new format. The bits will be recorded using a range of storage technologies, including recordable miniDVD discs, solid-state memory cards and hard disk drives.

At the other end of the quality/price spectrum, Panasonic has announced a range of professional products using AVC-Intra compression. AVC-Intra is said to be twice as efficient as the Panasonic DVCPRO HD codec.

The Panasonic AVC-Intra implementation of H.264 is flexible to user demands by virtue of its ability to switch between the AVC-Intra 100Mb/s mode and the more economical AVC-I 50MB/s mode. The AVC-I 100 mode provides full-resolution (no subsampling) HD. Both codec implementations are 10 bit. It is likely that there will be many other AVC product introductions at NAB2007 with the potential to offer high-quality recording of 10-bit 4:4:4 RGB for broadcast and digital cinema applications.

Craig Birkmaier is a technology consultant at Pcube Labs, and he hosts and moderates the OpenDTV forum.

Web links

Send questions and comments