SeaWell Launches Lumen1000 Appliance
At the Streaming Media East 2010 show, held today and tomorrow at the New York Hilton, SeaWell Networks announced the release of its Lumen1000 SVC-based video transcoder.
SVC, or Scalable Video Coding, is a method of transcoding a file in one format into a single H.264 file that has up to 27 different versions-in terms of bitrate and resolution-all packaged within a single file.
In some ways competitive, but in other ways complementary, to adaptive bitrate (ABR) streaming, SVC also uses a "chunking" approach, where content is divided into multi-second chunks at varying bitrates. When delivery network bitrates fluctuate, content shifts up or down to the next higher or lower resolution or frame rate.
“SVC can offer significant savings over existing technologies by managing complexity and cost,” said Colin Dixon, senior partner at The Diffusion Group. “Now a single stream can be produced that is easier to create, manage, and store and is optimized for a wider variety of devices and connection speeds.”
SVC requires the player to be aware of the variety of stream quality options, so that it can choose the best chunk at any given point. It also requires an SVC codec to encode content at the various resolution or bitrate slices, which is where SeaWell positions itself.
While the company is primarily a software company, the need to have a demonstration device that showed off the power of SVC encoding is key to the release of Lumen1000.
The Lumen1000 handles video transcoding from MPEG-2 or H.264 files in AVI, FLV, MOV, MP4 and a few other container formats. Using two Gigabit Ethernet connectors, and a workflow-based graphical user interface (GUI) that steps through transcoding and delivery, the device has a short enough depth to sit in a 14-inch deep AV or studio rack. Yet the web interface allows for connectivity from a traditional network data rack.
In a demo of the device the day before Streaming Media East launched, the company showed a workflow-based graphical user interface (GUI) that steps through transcoding and delivery. Choosing the setting step of the workflow revealed a series of options that can be tweaked to change the number of slices, length of chunks, and other key tweaks.
The user interface interacts to allow for a 3 by 3 by 3 selection of Spatial, Temporal, and Quality settings to get to the 27 possible slices. I asked Andy Beach, chief evangelist of SeaWell Networks, why these three were chosen and whether they are the most important elements for stratification when a compressionist chooses to slice the content into various bitrates and resolutions.
"These three dimensions have the greatest impact on supporting the application for scalability," says Beach, "including the ability to adjust to bandwidth, device type and processing power. They are the most efficient to support scaling (i.e. encoding and decoding complexity) when it comes to basic SVC settings."
The language of selectable options in the user interface (including Group of Picture (GOP) settings, chunks sizes, etc) appeared quite familiar, so I asked Beach whether the language was intentional as a way to explain to those familiar with ABR or traditional GOP solutions how to best choose SVC options?
"Yes, exactly," says Beach. "Truth be told, the selection of these parameters have very little association to SVC, but are the parameters of the existing ABR solutions. We provide some additional flexibility to allow a chunk size to be multiple GOPs if the encoding user so desires."
So why, I wondered, have both the Spatial, Temporal, and Quality triumvirate and the choosing of GOP and chunk size?
"For ABR solutions, including SVC, there are two fundamental areas of configuration that have impact on two different components of the solution," says Beach. "The first is scalability parameters - determining how dynamic the adaptive solution is, such as how many bit rates can be supported, how many different devices will be supported, and the like. In essence this replaces the need to create multiple bit rate files as these are the same base parameters that are used by existing ABR solutions."
"The second area of configuration is changing the delivery parameters of GOP size and chunk size," he says. "These parameters are more specific to the type of network that is used for delivery. For certain networks it may be better to have smaller chunk sizes so that the solution can adapt more quickly and frequently. If the chunk sizes are too small, however, it requires a lot of requests and impacts the performance of the servers delivering the video."