Prime Image introduced a new machine-learning AI capability, powered by its patented Video Content Analysis engine that automates the detection and recognition of branded objects in video content. Rich and extensive content metadata dramatically improves the ability for content owners and their brand sponsors to provide a rich viewing experience while also driving significantly increased revenue from their content monetization initiatives. Prime Image is using deep learning methods to compile and train its models, ultimately yielding higher classifier accuracies.
The costs associated with Streaming Video distribution are a huge component of operational costs and continue to grow exponentially. While various technologies exist to optimize bandwidth consumption, no one has exploited the opportunity to optimize content run-time. Prime Image has a cost-effective solution to deliver incremental savings above and beyond bandwidth optimization by reducing content run-time by 10% or more using standard H.264/265 codecs in the mp4 container. This reduction can equate to more than a 10% savings in the number of bytes streamed to end-users using proxy media files. This capability can save content owners, distributors and CDNs millions of dollars, and improve the user experience by mitigating buffering and network latency.
Prime Image has the ability to deliver multiple video content essence files (video and audio) where one video is considered the master and matches a chapter list that has been defined. Chapter points (or generic spans) fall on edit points in the program. A new chapter list is created for each of the additional versions. This process assumes main body of content is the same but that the head of the file has changes in some way due to platform delivery header requirements. The automated retiming process realigns chapter points using frames and TC information from the master chapter list. This automated process can be applied to any number of derivative assets per their required distribution and time clock.
Our Black Slug Reformatting capability allows a customer to define a set of rules against black content found in a video program and create a new output file based on the desired processing. This can also be removing all black from original commercial breaks for OTT streaming. This can also include a reverse function combining of the chaptering where a user might want to insert n seconds of black at each defined chapter.