How to Convert DVDs for Adaptive Streaming: Part 1 of 4
Recently, I participated in a project that involved converting several DVDs to multiple-file, adaptive-streaming output. This project involved several interesting technical issues that I'll present in this multiple part case study, including how to rip the DVDs (which were not copy protected), which video editor to use to sequence the imported files, which sequence preset to use during editing, where to deinterlace the video, how many streams to produce, and at what resolution and data rate. How to test the output was also an interesting issue.
I'm going to break this case study into four sections, presented a week apart. In this first piece, I'll introduce you to the players and the project, and discuss the decisions that had been made prior to my signing on.
In the second section, I'll detail how we came up with the number of files to offer in the adaptive streaming offering, and their configurations, and how we tested the multiple file offering.
In the third section, I'll discuss the workflow used to rip the files from the DVDs and edit them in in Adobe Premiere Pro. This includes how I created the sequence presets that I used, the filtering that I applied, the format that I used to create the intermediate files, and why I decided to deinterlace and render to final format in Sorenson Squeeze, rather than Adobe Premiere Pro.
In the fourth section, I'll detail the procedures used to produce the final files in Sorenson Squeeze and the quality control procedures that I used to test the files.
The client was a company named SunCam
, which provides continuing educational courses for engineers, architects, contractors, and project managers. Many of its courses originated on DVDs, and the specific courses involved in this project were three DVDs from The Behrend College of Penn State University, Erie Campus.
I was brought into the project by Stefan Richter who would handle the programming side. My job was to convert the video from DVD to the appropriate streaming format and suggest the optimal number of streams and the appropriate resolutions, data rates, and other relevant configuration options.
My contact was SunCam CEO William C. Dunn (Bill). In our early discussions, Bill let us know that he had considered streaming before, but that the quality had never been sufficient. The classes are all long form (the DVDs averaged about three hours of content), so Bill felt strongly that screen size had to be as large as possible, and the quality very close to the original DVDs. In this regard, Bill commented several times during our initial discussions that he hoped to achieve quality consistent with the NetFlix videos he watched at home. He made it clear that if we couldn't meet that goal, he likely would shelve the project until that quality level was achievable.
By the time I came into the project, several key options were already decided, most importantly that SunCam would use Adobe Flash-based Dynamic Streaming powered by the Amazon Web Services S3/Cloudfront. From there, it was a pretty simple decision to use the H.264 video codec. I asked about distributing to Apple iPhone and iPad devices, which the company decided to defer, primarily because the duration of the video seemed to make it likely that most viewers would watch on computers rather than smaller screen devices.
For you Flash coders out there, note that Stefan wrote a blog post documenting his experience delivering adaptively via CloudFront.
That's the background; next time around we'll look at how we figured out how many streams we decided to include and how to configure them.
Jan Ozer's article first appeared on OnlineVideo.net