Edgar Huang, Ph.D. School of Informatics
Clifford C. Marsiglio, Testing Center
Indiana University–Purdue University Indianapolis
Discussions and Conclusions
Here are the itemized analyses of each HD video streaming technology based on the findings from this study.
Adobe Flash HD video is probably the biggest winner in this test. It excels consistently regardless of bandwidth. The almost ubiquitous pre-installation of Flash Player in all Web browsers, its outstanding image quality, almost instant showing, together with the full-screen button (enabled by ActionScript) newly built into some of the skins, make viewing of Flash HD streaming video a truly hassle-free and enjoyable experience to the testers. Flash HD video is the closest to owning the capability of presenting both high streaming quality and high image quality simultaneously. In early 2008, Adobe reduced the cost of newly introduced Flash Media Server 3 to $995. This move may encourage more companies, organizations, government agencies and schools to stream Flash videos by themselves rather than pay exorbitant fees to a CDN or live with the bad quality of Youtube videos. The obvious downside of using Flash video is the Flash CS3 Video Encoder’s slow encoding time and the inconvenience of encoding multiple bit rates in comparison to some other technologies, such as DivX, QuickTime and Windows Media. Compared to QuickTime, Flash video has less rich colors, as many testers pointed out. The full-screen feature should have been built either into the Adobe Flash Video Encoder or Flash so that no extra ActionScript writing is needed. In the introduction, we set the bar for an ideal HD video streaming technology as follows: "An ideal online HD video streaming technology is one that presents HD image quality without long initial buffering or any rebuffering. It should also be easy, quick and inexpensive to encode such a video at as small a file size as possible. Finally, it should have an easy full-screen support." Adobe Flash HD video is by far probably the closest to such an ideal.
DivX HD video is another big winner in this tesing. Probably the best word to describe DivX HD video technology is “handy.” First, the whole DivX Pro suite for streaming HD video is only $19.99. It is the least expensive of all the tested technologies. Almost anyone can afford it for presenting HD videos online. Second, it encodes HD videos the fastest among all the tested technologies, and it also does two-pass encoding. Third, it generates one of the smallest file sizes. However, the small file size can be a double-edged sword. DivX has the least desirable image quality. In comparison to the image quality of Vividas HD video, which generates the same small file size, the DivX counterpart has much to improve. The soft and pixelated video image, to some extent, defeats the purpose of presenting DivX HD video, which is supposed to be crisp. Fourth, when viewers watch a DivX HD video, they experience the least rebuffering, and the video also pops up quickly within five seconds for most Internet connections. By double-clicking a video embedded in a Web browser, a user can watch the video full-screen. No other technology can do that trick. Finally, DivX HD video does not involve a streaming server. Rather, DivX shows a movie via progressive downloading through DivX Web Player. All video files are stored on a Web server. Strictly speaking, this technology is not a streaming technology, but its ease of use, inexpensiveness and DRM feature make it a viable alternative to HD video streaming. The biggest drawback of using DivX HD video is its extremely low pre-installation rate because it is a new player on the block. The request for installing unfamiliar software may scare away some users. Overall, this handy technology is suitable for those who like to post HD videos online quickly but can live with DivX's mediocre though still acceptable image quality.
Apple QuickTime HD video presents the best image quality among its peers by a huge margin. This HD technology is inexpensive and easy to learn. Its fast encoding can really make a media producer happy. Mac users can have a very pleasant viewing experience of QuickTime HD videos as many Mac users in this test have reported. On the other hand, since QuickTime is not installed by default on some PCs, some users may refuse to watch a QuickTime movie when asked to install QuickTime. Windows Media Player cannot even recognize a QuickTime movie. Also, the performance of QuickTime on a PC is not as stable as it is on a Mac. On average, it takes three more seconds to launch the same video in the PC-version QuickTime than in the Mac counterpart. Another downside is that QuickTime does not have a full-screen button built in its controller when a QuickTime video is embedded in a Web browser. Therefore, the viewing of a QuickTime HD video has to be assigned to an external QuickTime Player. The triggering of an external QuickTime Player on a PC is tricky. In our experiment, after clicking an image link, an external QuickTime Player does not automatically come up. Instead, a user has to click the Web page that contains the image link to trigger the player. Ordinary users may get frustrated with such a setup. Overall, PC users may not be happy with their experience of watching a QuickTime movie. Considering the fact that there are still drastically more PC users than Mac users in this world, QuickTime HD videos will probably continue to be a niche hobby than a commercially viable option for HD video producers.
Vividas HD video is a Java-based streaming technology. Like the VX30 technology we tested in our 2006 on-demand streaming video study, Vividas video streams through the HTTP protocol. No streaming server is involved. It does not rely on any particular media player, such as QuickTime, Windows Media Player or Flash Player. When a video is being played, a Java applet is sent to the user’s computer. Vividas HD video has quite decent image quality. With the appropriate amount of initial buffering set up, the playing of a Vividas HD video should not show any rebuffering. Convenient multi-bit-rate encoding, batch processing and DRM are all good features for media producers. On the other hand, this promising technology still has room for improvement. For instance, the long initial buffering really tests users’ patience; quite several testers reported that Vividas HD video crashed their Web browsers; scrubbing video is not possible because the controller does not contain such a feature; video encoding takes a much longer time than it does for any other tested technology. The exorbitant prices that Vividas charges could be a drawback for Vividas technology to be widely adopted. Finally, as of the time of this writing, Vividas On-demand Encoder can encode up to 2500Kbps for videos. Higher bit rate video encoding is impossible without Vividas personnel’s help.
Microsoft Windows Media HD video is very easy and comparatively fast to encode especially when multiple bit rates are needed. However, this technology has a mediocre performance on almost every count. It has also suffered from its discontinued support for Windows Media Player on Mac. Not all Mac users have the required Flip4Mac installed on their computers to watch Windows Media videos. In addition, if a user has VLC installed, VLC may pop up to play Windows Media videos, but not smoothly. Testers with a lower-than-5000Kbps broadband connection experienced extensively more rebuffering. An embedded Windows Media Player cannot show a video full-screen. As a result, an external player has to be used. Although Windows Media Encoder is free, the Windows Media Server 2003 or 2008 is very expensive. While Adobe Flash Media Server 3, which enables almost guaranteed streaming performance, needs a much smaller investment, customers must have a very special reason to pay much more to lose potentially a large chunk of users. Informed customers may make their decisions based on the fact that MSNBC, a Microsoft company, even uses Adobe Flash video, not Windows Media Video, to stream videos.
Acknowledgement: The author would like to thank Dan Gregory and Todd Kirk for their generous technical support for the study.