The idea was bold, masterful, and instantly appealing. Take a classic song from an iconic artist—Bob Dylan’s Like a Rolling Stone—and create 15 additional versions lip-synched by stars from multiple TV channels. Display them in a TV-like player that lets the viewer flip from channel to channel with a continuous audio soundtrack. Click, Drew Carey singing on The Price is Right. Click, Steve Levy singing on ESPN. Click, one of the Property brothers singing into his cell phone. Click, back to Dylan, where it all started.
All the channels are synchronized, so you can watch the song from start to finish looking at any or all of the sixteen performers. It is an irresistible, captivating presentation that makes you want to play the song again and again.
Finding a medium that would make viewers want to play a song multiple times was the entire point of the technology behind the Dylan video, which was produced by Interlude, an Israeli startup company. Interlude was co-founded by Yoni Bloch, a popular musician, who also starred as a judge in the Israeli version of American Idol. As Bloch explained to TechCrunch in 2013, “In the music industry, they say if someone hears a song three times it will stick. So we wanted to make an interactive video that would make the song stick. But we couldn’t find any technology that did it the way we imagined it was supposed to be, so we built it ourselves.”
According to Bloch, Interlude started selling the concept and productions to other artists. Soon major brands started calling, lured by the higher retention rates, deeper engagement levels and higher percentages of sharing enjoyed by Interlude’s interactive productions. As an example, according to the Times of Israel, in the week following the release of the video on BobDylan.com, traffic on the site jumped tenfold.
Build It In Treehouse
All Interlude productions are created in Treehouse, the company’s authoring suite. “Though Treehouse simplifies the creation and design of interactive productions, what goes on behind the scenes is anything but. Since we’re not delivering a linear video experience, our storage and mainly delivery volumes are multiplied by a factor that is tied to the creative—the more rich and elaborate an interactive video created with the platform is—the bigger the underlying structure is, and as a result the amount of video segments to be stored and delivered.” Explained Tal Zubalsky, Interlude’s co-founder & CTO.
As an example, to enable the highly-responsive channel switching in the Dylan video, Interlude plays the video from the selected channel, and pre-loads segments from the two channels adjacent to the selected channel, one upwards on the channel selector, and one downwards. This makes channel switching instantaneous, but it also forces Interlude to deliver three videos simultaneously.
Each Interlude project is different, of course, and as Treehouse gained more creative options, Interlude has implemented a range of new techniques to support the required interactivity and responsiveness. Given that Interlude is working with video, an inherently bulky medium, this created a consistent set of stringent video file-related requirements. According to Zubalsky, “we need to ensure good video quality and fast loading times to our viewers, while keeping buffering rates to a minimum.” To meet this challenge, Interlude turned to Beamr, for its video optimization solution – Beamr Video.
Optimized by Beamr Video
Beamr Video is a perceptual video optimizer that significantly reduces the bitrate of video streams while preserving their full resolution and quality. This reduces storage and bandwidth costs, and enables the delivery of higher quality video over low bitrate connection.
The Beamr Video optimization process is a closed loop system that guarantees the preservation of subjective quality when reducing video bitrates. It is based on a patent-pending perceptual image quality measure, developed over 4 years of intensive research. Beamr Video processing is fully automatic—there are no configuration options, just two settings, Best and High.
Beamr Video runs on 64-bit Ubuntu 12.04 and RHEL 6.x and compatible systems, and is applied to encoded MP4/H.264 files before encryption and packaging, or to H.264 elementary streams targeted for Blu-ray production.
Beamr Video—A No Brainer
Interlude uses Beamr Video optimization for premium clients that entail projects with obvious mass-market potential, and videos that have garnered sufficient viewers.
Interlude has Beamr Video running on several dedicated servers sitting close to the professional services and creative teams. Once the company decides that a particular project will be optimized, it generates an automation script that fetches a project’s video segments, optimizes them and uploads them back to the CDN.
According to Zubalsky, Interlude runs Beamr Video in the Best quality mode, which proved to be near lossless in terms of quality, though it provided less optimization than the High quality mode. In addition, Zubalsky commented that videos with slow movement and fewer cuts provided greater optimization gains than fast moving videos with lots of cuts.
Not surprisingly, for a company producing highly interactive videos for online delivery, the key benefit of Beamr optimization was related to the user experience. “The main benefit for us is faster loading times and less buffering for people watching Interlude videos,” stated Zubalsky. “We didn’t A/B test projects to see loading times or buffering rates on the same project with/without Beamr, but with the percentage of reduction in file size-it’s a no-brainer.”
About Beamr
Beamr is the global leader in media optimization solutions, powering some of the world’s top web publishers, social networks and media companies. Beamr offers a patent-pending perceptual video optimizer, which reduces the bitrate of H.264 and HEVC streams by up to 50%, preserving their full resolution and quality. By reducing video bitrates, Beamr enables content and service providers to distribute exceptionally high-quality video, with faster downloads and smoother streaming
on bandwidth constrained connections.