Rally for Content Logging
Content is logged for a number of different reasons, but the end result is typically the same: an accumulation of time-based metadata pointing to a specific location (or time range) in the content identifying a particular “event” on the timeline.
Logging is used to identify a wide variety of “events”:
Possible ad breaks
Clock/ bars/ tone
Sections of black signifying break points
Content inappropriate for certain audiences or markets
Tagging themes, objects, or categories to simplify location in the future
Traditionally vs. Today
Content logging has been a labor intensive and repetitive task, often performed manually by specialized staff. Data logged on different types of events is difficult to share across teams and operational groups. Quite often the efforts of multiple teams will overlap, wasting valuable time. One team may identify all the segments and black for a particular distribution version, while another team does a separate review the media to look for inappropriate content. The resulting bottleneck in media operations can limit overall capacity so severely that it will kill distribution deals.
The Rally Media Supply Chain Platform provides tools to dramatically overhaul logging. AI and ML data available from a range of 3rd-party services identifies and flags events for operator review, increasing efficiency, allowing for greater throughput and improving business agility. Once your content is cloud resident, leveraging one of the cloud-native computer image services is a straightforward task.
Process proxy through AI services
Normalize time-based metadata
Review events/ labels in context of the content
Smart Utilization of Proxies
It’s typically more efficient (faster, cheaper, and just as good) to perform such operations on a decent quality proxy version of the content rather than the master. When instructed, or automatically based on the completion of a previous gate (such as content receipt/ link), Rally can automatically make a proxy version of your content. This proxy can be used for a number of other tasks within the supply chain such as browse, QC review, review and approval, but it can also be used for logging.
Once the proxy is available, Rally can instruct an AI service to analyze that content. It’s possible to limit the class of things you are looking for, but it’s typically better to capture it all, and simply filter the data later based on what you are looking for. The collection of AI services can return everything from object labels, adult content filtering, black detection, scene change detection, on-screen character recognition, transcriptions and more. Once within Rally, this data can be normalized along with any other time-based metadata (such as QC events or issues) and made available to operators in tools that will help them complete their task.
Rally Gateways (l to r): review & approval; status dashboards; content & metadata delivery
Streamlining the Work of Operators
The Rally platform provides a range of ways that operators may utilize this data to ease their task. Firstly, any selection of that data can be made available via both the metadata and player widgets in Rally Gateway, enabling operators to quickly navigate to, review and determine any actions. While there, they are also able to adjust the handles on each event and add their own manual events.
Rally Access provides a more comprehensive, edit-friendly environment for both review and fixing of any identified issues. Is the task is to review content for appropriateness for distribution to a specific territory, Rally can filter the available data to only show issues relevant to that task. Users have to simply navigate the highlighted clips, determine how they want to fix them (remove, obscure, dip audio etc.), and save the workorder. Rally then saves all the metadata changes back to the content record, and kicks off a remote conform process on the edited content.
Rally customers have realized significant benefits from this approach, from savings on traditional edit suites used for such tasks, restructuring of shifts as operators can now process more content in less time, and the ability to sell more content through electronic sell-through channels due to the ability to process more content faster.
AWS Rekognition · AWS Transcribe
Google Video Intelligence · FFmpeg · Piksel Fuse