How to Increase Revenue by Monetizing Video Content with Time Tailor and AWS
By Bennett Lomax, President – Prime Image
By Greg Jankoski, ISV Growth PDM, Emerging Technology Partners – AWS
As media companies expand free ad supported television (FAST) channel delivery and explore other advertising-based video on demand (AVOD) possibilities where they may have multiple format demands for various channels, Prime Image’s Time Tailor saves them significant cost and time.
For many years, linear television networks have used Time Tailor in its prior hardware form to customize content to maximize revenue and improve the viewing experience. Historically, this was a “back-of-the-house” editing service that network operators deployed to control their clock. These edit services were expensive, subjective, and took days to complete.
Prime Image is an AWS Partner and AWS Marketplace Seller that pioneered automated micro-editing with its patented Time Tailor products and artificial intelligence (AI) services to customize content runtime across media management workflows.
Time Tailor’s patented algorithms analyze discrete video pixels and audio samples through a dynamic edit decision framework. This method, combined with dynamic interpolation, delivers media run time with no artifacts, change in lip synch, pitch, or tempo across 16 channels of audio. With this quality, Time Tailor became a well-known industry secret used by every major North American television network, content producer, and cable operator.
Time Tailor has been successfully converted to software and deployed across some of the world’s largest media companies. Leveraging the scalable compute infrastructure of Amazon Web Services (AWS), Time Tailor runs as a cloud-native application that can automatically re-time content up or down to meet customer broadcast and streaming requirements.
Combined with AWS’s elastic cloud infrastructure, Time Tailor can process content faster than real-time, allowing last-second content customization without long lead times and costly editing.
In this post, we will discuss some of the use cases for Time Tailor in different organizations, where it fits into the media supply chain, and the underlying technology.
Managing Content Ad Inventory
With the growing adoption of IP video and networked computing infrastructure, media companies now have (a) increased flexibility to control production and post-production activities, and (b) new avenues of distribution for their content. Many of these new outlets have different versioning requirements, and content needs to be positioned for different advertising inventory loads and must be compliant in different markets.
In addition to cable and satellite television distribution, the media industry is rapidly evolving to include linear streaming channels where advertising is the primary revenue driver. There are over 1,400 new FAST channels and the market is growing.
In the current environment, where content owners are seeking new ways to monetize their content, the need to repurpose (re-version) existing libraries is likewise growing. Time Tailor and AWS deliver micro-editing to help re-version libraries and position content for new revenue opportunities.
For FAST channels, Time Tailor can seamlessly remove ad breaks from content that was originally produced for television and is now being repurposed for streaming.
Time Tailor can re-version content with different ad loads for different distribution channels. Each distribution outlet has different allowable ad loads, and Time Tailor provides a cost-effective, timely process to adjust content for additional ad opportunities.
Time Tailor’s automation-first framework was originally designed to manage media assets, allowing media operators to queue jobs and dynamically and accurately edit broadcast programming with little to no human interaction.
The first generation of this technology was deployed 25 years ago, when Prime Image changed the way the television industry modifies video and audio content with Time Tailor, an editing automation appliance used by broadcasters and cable networks.
Time Tailor allowed editors to scale their output inside of a network operations center by managing the length of their linear programming to meet desired clock times, without a human having to make hard cut edit decisions.
Whether with live or recorded programs, Time Tailor became an indispensable tool inside of major North American broadcast networks, helping them automate editing functions with no degradation to content quality.
Prime Image developed a proprietary media asset pipeline leveraging patented IP around media asset clock management. The pipeline is fully extensible using a proprietary automation framework built on both Linux Ubuntu and Windows Server 2016.
The web container is Glassfish and the framework utilizes related tools and rest standards to support and manage all API transactions. The pipeline includes a software transcoder that supports bi-directional asset transcoding, decoding, and encoding of standard broadcast and over-the-top (OTT) formats.
Decoding of the media stream happens in memory to reduce processing latency, specifically for jobs that don’t require new video outputs. The automation framework works in both frames and time code (TC), which enables customers to build edit decision lists (EDLs) or equivalent job descriptors into segments, chapters, or total frame counts.
The pipeline’s core model is centered around the OpenCV library. This tool provides the construct for frame inspection down to polygons or blocks of pixels, coupled with Prime Image’s proprietary algorithms to measure motion, object vectors, and pixel similarity.
The core model also includes an audio processor that analyzes every audio sample in 512 blocks and runs a proprietary waveform algorithm to identify matches, followed by splicing to remove samples and re-blends the audio stream. This process is not detectable during playback and does not require pitch correction.
Utilizing the pipeline and native AI functionality, Prime Image addresses the need to disintermediate the highly disparate AI media service industry through a single input/output interface and pre-classifying job type to optimize AI results.
Prime Image’s patented vertical cut (VC) detector is a key differentiator in the AI processing sequence. The pipeline auto-runs the VC detector on all media assets, and this metadata provides frame sequences by camera and is used to capture specific frames within the sequences to be passed to AI services. This reduces the number of frames and processing time by media asset.
Figure 1 – Time Tailor solution.
AI Pipeline Service Description
Prime Image developed a standalone AI pipeline and service controller to support native AI model processing and further mediate input compliance to third-party AI model partners.
This service interface leverages the media asset management pipeline and inherent filters and analyzers to improve and optimize AI processing intelligence, resulting in fewer false positives and reduced processing costs by job request. The AI pipeline contains Prime Image models, along with partner models that can be added to this extensible framework.
Prime Image’s frame similarity work process provides a unique capability to quickly create actionable hotspots in media assets, thereby avoiding the more traditional model-based training approach. The technology enables content creators to develop an immersive user experience by targeting interactive hot spots or deleting undesirable content from the media asset.
The process targets the frames or images a user cares most about within the media asset, from product placement to content moderation, pattern recognition, and product merchandizing. This advanced method of frame and image matching is used as a pre-filter to target and isolate frames within the frame similarity process.
The frame similarity detector uses a multi-step iterative detection process to narrow down scenes, frames, or TC that contain target images or frame content. The image below provides an overview of the pipeline data services.
Figure 2 – Time Tailor + AWS workflow.
AI Pipeline Scalability
The AI pipeline processes operate within the CPU and utilize cores and processor clock speeds to drive throughput. To improve native performance and throughput, much of the processing is managed in memory using a complex buffering architecture. This approach eliminates disk I/O latency and generates quicker transcode times to and from the analysis engine.
Throughput can vary by 50% depending on the type of job (detection vs. alteration). Other alteration or content moderation jobs where the process generates a new output file can take up to 1.1x of real-time of the processed asset. This process generates encoded mxf, mov, and mp4 container files at the customers desired bit rate.
Prime Image’s AI pipeline application runs in a standard Docker container. The container can run on any host (node, virtual machine) or cluster using native cloud architectures and elastic services, such as Flexible App Engine and Amazon Elastic Container Service (Amazon ECS).
This service, with Elastic Load Balancing, enables the pipeline to elastically shape capacity with demand, while maintaining 100% utilization and price-performance from the job queue manager.
The new software-as-a-service (SaaS) Time Tailor improves an existing industry standard with new technical and commercial benefits.
- No up-front costs: Time Tailor is an on-demand service with a pay-as-you-go pricing model where customers pay only for what they’re using. No upfront costs; no capex.
- Easy installation: All you need is a designated Amazon Simple Storage Service (Amazon S3) folder and Time Tailor will show you how to process your content at your desired duration.
- Dynamic scalability: Time Tailor is dynamically scalable as it runs on Amazon Elastic Compute Cloud (Amazon EC2) without the use of costly graphics processing unit (GPU) resources. Large video processing batches can be queued to run in series or in parallel depending on customer needs.
- Custom workflow integrations: Time Tailor has been integrated into some of the world’s largest and most advanced media operations. It can plug in to third-party media orchestration platforms or embed as a customized microservice directly into any customer’s broadcast infrastructure.
- High-touch service: Time Tailor includes a standard support package that delivers Prime Image’s highly responsive technical support. The support experience can also be customized through premium support options.
Prime Image – AWS Partner Spotlight
Prime Image is an AWS Partner that pioneered automated micro-editing with its patented Time Tailor products and AI services to customize content runtime across media management workflows.