Streams of Thought: H.264 Wins—Now What?

This column originally appeared in the European edition of Streaming Media magazine. Click here for subscription information.

H.264 set the stage for dominance a few years ago for a common live, on-demand, and broadcast format. Compared to MPEG-2 (the format of traditional digital television and DVDs), H.264 offers two to three times greater compression, making it much more attractive for network delivery as well as for HD video.

The need was simple: a common decoding format that could be used across the streaming, videoconferencing, and IPTV segments. H.264 enables content created for one type of device to be easily delivered or adapted to another, at least in theory. The European-embraced ideal that a standardised open format drives competition and reduces the cost of devices, thereby expanding the addressable market, also means that devices as dissimilar as home computers (Windows, Macintosh, and Linux), the leading mobile devices (iPods and iPhones from Apple or handsets from Nokia and Sony Ericsson), and DVRs and IP set-top boxes all use H.264 at their core.

Also, with a common format consumers and businesses are encouraged to create and share more media, as they know that broad distribution is possible. Because of this, consumers, media companies, and vendors alike all benefit from increased growth, innovation, and choice.

True, there are still other formats out there: Microsoft has VC-1 and On2 Technologies has VP6. But even these two powerhouses are embracing H.264: Microsoft’s IIS 7 server component supports H.264, and it has just announced it will demonstrate H.264 in Silverlight later this week at the International Broadcasting Conference (IBC) 2008 in Amsterdam. It will be available in 2009. For its part, On2 owns Hantro, a European company that uses H.264 for embedded video delivery. There are more announcements to come from both of these players, one suspects, as the H.264 juggernaut continues to roll forward.

Now what? Will the streaming, broadcast, videoconferencing, and on-demand video worlds all suddenly stop innovating, having been conquered by H.264, an analogue of Alexander? Not exactly. We are entering another round of Pax Romana (to mix metaphors), yet big challenges lie on two fronts. First, the player. This is a large issue, as the format wars have given way to a player war. Since H.264 can be encoded in any way a developer sees fit but must meet a particular decode standard to qualify as H.264 video, the decoders in players are very important.

Consider, for instance, Sanyo’s H.264 camera. The Xacti HD1000 camcorder is an H.264 1080i camera that captures full 1920x1080 (1080i) at 60 frames per second using a newer CMOS sensor. It captures on an SD or SD High Capacity (SDHC) chip and looks rather good. In tests with this camera right after its release, the files could be dropped into QuickTime Player for immediate playback, as they were .mp4 files.

Unfortunately for Sanyo and its customers, Apple "fixed" something in QuickTime that suddenly made all video clips shot on the Xacti HD 1000 unreadable in QuickTime Player, iTunes, and almost every other third-party application that relied on the underlying QuickTime engine. This is just one example of what’s happened several times across both the Macintosh and Windows platforms as H.264 encodes are written not to spec but to fit a particular player.

That issue, writing H.264 to spec, is the second—and bigger—challenge we now face.

Often, as was the case with AVCHD, the format isn’t quite up to H.264 spec, so it’s couched in terms like "based on H.264" or other marketing speak. AVCHD is a format that was jointly created by JVC and Sony in an attempt to get a tapeless format that is approximately 25% better than HDV in terms of recording bitrate. Unfortunately, AVCHD isn’t H.264 and, as such, requires conversion into another format for video editing. The conversion that Apple does creates image quality loss, and Adobe has yet, as of the time of this writing, to come out with a native AVCHD solution (the third-party solutions they recommend are only single-platform fixes).

Enterprising companies such as Blackmagic Design have figured out that they can capture directly off the HDMI output from an AVCHD camera and encode it to another format, but that in itself creates issues for editing (no timecode via HDMI) or streaming (an uncompressed stream that needs to be compressed).