NMC 2007 - The Ideal Stream

The final session I attended during the Attended 2007 NMC Conference was presented by a group from the host institution. Edgar Huang and Clifford Marsiglio of the School of Informatics at Indiana University Purdue University Indianapolis (IUPUI) and their team have been conducting a methodical survey of available streaming technologies and their capabilities. The extensive results haven't been published yet, but the conclusions they shared are, in some cases, surprising.

The study, Searching for an Ideal Live Video Streaming Technology considers eight formats: Clipstream, DVTS, Flash, Quicktime, Real, VLC, VX30 and Windows Media. Not surprisingly, the I2 formats (DVTS and VLC) rate among the best for quality of image and minimum of latency. It was the Flash codec that really stood out, however (and, again, unsurprisingly). When the technologies were ranked by user experience (cost, ease of use, accessibility, etc.) it seems that it's tough to beat the ubiquitous Flash engine. It's by no means the best codec by an image quality standard, but it's the one most likely to play reliably on any PC on the net - and the image is good enough.

As I said, the study hasn't been released yet, but an earlier version of the project from last year is online: Searching for an Ideal Streaming Technology .

Related Blog Entries

Comments (Comment Moderation is enabled. Your comment will not appear until approved.)
When commenting on image qualit in streaming technologies, you really need to qualify that with some bitrate information. I would hazard a guess that the On2-based Flash video codec looks as good or better than most others at very high bitrates. Do you mean that Flash video compares poorly in image quality at low bitrates? What bitrates are you talking about? It is nearly meaningless to state Flash is not the best codec in image quality if you don't qualify that with information about which bitrate you are comparing the codecs at.

I'd like to see what the differences are. Several of the items you list above are also ambiguous, because they are wrappers which can contain more than one codec for video compression. MPEG4 variants might all have similar compression artifacts, for example with high motion in video. Where those artifacts appear and to what degree might vary depending on artifacts. Compression settings tailored for different kinds of video will greatly effect image quality. i'm interested to see the results of this study, especially their methodology. AtomicPC recently did a nice review of codecs (though not focused on streaming):

http://www.atomicmpc.com.au/article.asp?CIID=82898...
# Posted By Clay McGovern | 6/19/07 9:39 AM
Consider my post simply an overview. Bitrate is, as you say, very important, especially as you try to compare the apples and oranges represented by the various technologies. The researchers absolutely accounted for these differences. Although the most current study has yet to be published, you can get a sense of how they operated from reading about the methodologies used in the similar study from last year: http://www.iupui.edu/~nmstream/study/methodology.p...

Take a look and you'll see how bitrate figured into the study and the results.
# Posted By Derek | 6/19/07 11:16 AM
BlogCFC was created by Raymond Camden. This blog is running version 5.8.001.