Virtual Video Clips showcase: Adaptive Streaming

back

Adaptive streaming is similar to pseudo streaming, but instead of serving one large video file, the video is delivered in a series of short HTTP downloads (segments).

This delivery technique is described in this article. Note that we are requesting Virtual Video Clips from the webserver. The underlying video is just *one* file, so you can use your original encoded videos. There is no need to transcode them again or cut them in hundreds of small files.

The pros of this method are:

• Fast start: an initial burst is sent to quickly start the movie.
• Bandwidth shaping: it only delivers the parts of the video that the user actually watches.
• Seeking: jumping to a not yet downloaded part of the movie is fast.
• Dynamic: the client can dynamically adjust to bandwidth availability and toggle between low and high video bitrates.

The cons of this method are:

• Although the streaming module delivers the segments accurate to the frame, the Flash player cannot append the video clips seamlessly.

This is an example of a Flash player using Virtual Video Clips for playback. The 1 hour and 50 minutes video below is partitioned into segments of about 30 seconds (aligned to keyframes).

The featured movie is Cowboy Canoe Coma.

Of course you can also encode the video at multiple bitrates and switch dynamically between them depending on available network bandwidth.

The example source code is available for review. Please also read the chapter Partitioning Multiple Segments in Lei Zhu's Efficient Video Delivery Over The Internet article.

If you have any questions, thoughts or ideas on Adaptive Streaming you can leave a message on our forum.