The first 2020 Tech Emmy award, “Development of Massive Processing Optimized Compression Technologies,” recognizes technologies that optimize video quality and file sizes by using the elasticity of the AWS Cloud to bring by bringing lots of computing resources to bear on each encoding job through parallelization. Adding more compute has always been one way to improve compression. In a traditional broadcast data center, this meant building a large physical fleet of transcoders, which meant spending lots of capital on equipment that might go underutilized most of the time. But, as many companies have migrated their computational and analytical workloads from traditional company-owned or shared data centers to cloud infrastructure, the transcoding and packaging of assets no longer have to compete for a set level of resources – the cloud makes it easy to scale up when you need to transcode a video, and then scale down when the job is complete. The result of this combination is reduced turnaround times in asset packaging and transcode, increased quality, reduced bandwidth, and lowered production costs. AWS won in this category for the software stack it innovated to encode and compress video materials, specifically the Accelerated Transcoding feature in the AWS Elemental MediaConvert service.
Accelerated Transcoding increases the processing speed of file-based video encoding jobs by up to 25 times, and is a feature in AWS Elemental Media Convert, a service that converts video into multiple output formats to support viewing from a broad array of devices at varying resolutions. For example, EPIX, a 24-hour premium TV network, now uses Accelerated Transcoding to convert 4K UHD content for streaming to approximately 70 million U.S. pay TV homes. With the addition of Accelerated Transcoding, customers can deliver more content faster via social media in parallel with their linear broadcasts.
However better real-time video compression isn’t just about more compute power; it’s also about being smarter, and that is where machine learning plays a role. One of the biggest challenges in video encoding is striking the optimal balance between compression efficiency and perceptual quality, or how good the video looks to the viewer. Intelligent content aware encoding configures the video encoder in real time according to the content of the video clip, allocating only the bits required for a given video based on its complexity.
