Lessons Learned from a SMF rollout

May 18, 2010 at 4:02 AM

I always find it useful when people follow up after successful projects with the 20/20 hindsight after running down the road a bit...

We just finished replacing our Windows Media infrastructure with a Smooth Streaming/CDN cloud platform product (Nimbus - http://www.nimbushd.com), and SMF has been our player of choice. We have used it for the most part out of the box aesthetically, with some customizations on the backend to facilitate our data services and realtime player analytics. We have used the timeline marker framework/pattern, smooth streaming (on demand & live), progressive download, but not the ad functionality (yet).

The SMF team has done an amazing job of creating a solid framework. I have been thoroughly impressed with the amount of thought that has gone into the architecture of this code. We really had little to solve with SMF, but here are the issues that we dealt with that could possibly be great in the next version of SMF, and I'd love to hear from others that have solved these problems also.

  1. Changing sources between Smooth Streaming and Progressive Download (detailed here - http://smf.codeplex.com/Thread/View.aspx?ThreadId=211178) presented some odd behavior that required a little creative code to solve.
  2. Non-Auto Start Player UI - When presenting a player on a page that is not supposed to Auto Start, the default black screen is somewhat unintuitive and confused users in our testing. We created a 'splash screen' with a video thumbnail & play button to solve this and mimic the experience of other similar players.
  3. "Go Live" UI - Still working on this; in our user tests, users aren't understanding what the "Go Live" button does (or why they're not "live" already). We haven't really come up with anything more smooth other than a "Jump to Real-Time" style change or a better way to represent a DVR.
  4. Video Thumbnail Generation - This was a pain point on the back-end; I'm sure this is outside the scope of this framework but it would make life easier if one could download an individual chunk and generate a thumbnail frame. We ended up using a big database of thumbnails at each 1-second interval in videos and integrating it into our initial encoding process; would be curious to know how others are doing this.

Thank you, SMF team for a great job and this highly extensible platform!