The digital video industry has major challenges with creation because editing and production costs are rising. Lack of video inventory is universal; however, there is hope in leveraging video artificial intelligence (VAI) to address these problems. Massive amounts of live and pre-recorded video never reach the audience. Publishers are being asked to do more with less. Publishers are hamstrung due to small editorial budgets and shifting priority while massive video revenue goes untapped. The need to intelligently create video inventory couldn’t come fast enough to feed the insatiable mobile video appetite. Here we dig in on the What, Where and Why in VAI and rethink how content is being created, edited and delivered all towards building greater video lifetime value.
Above video was created using video artificial intelligence. Animated preview thumbnail selected using machine learning.
Rise of Video Artificial Intelligence
Automation is not the cure-all; however, combining live streams and artificial intelligence solves several pain points while augmenting the production staff and delivering relevant content. Your editorial team is the most valuable resource in your video workflow. This is the very reason why these highly-valued editors should be focused on higher value video creation. VAI isn’t about replacing your editorial team but scaling lower priority content at a fraction of the cost while increasing revenue generation.
There are many promising video AI examples showing off real-time facial tracking, object recognition and image labeling etc. In our previous post Top Video Artificial Intelligence and Machine Learning at NAB 2018, we highlighted several early AI products in the market. Outside of the video labeling services several are extending existing video editing systems as a cloud Saas. So what does this all mean for my existing video assets? First off, there has been monumental advancements in algorithms to obtain a deeper understanding of actions people take within a video. Such as YOLO and Deep Structured Models to label individual actions in video. This in-depth dissection changes the game of measuring what parts of a video resonate with consumers vs. old linear video measurement which woefully lacks meaningful insights.
Promising Use Cases
Using VAI to create and edit video isn’t new. Companies Like Adobe and IBM are all using advanced video/image analysis to enable smarter editing platforms. However, publishers need more than just editing assistance they need scalable video creation. One of the more profitable use cases is short video highlights and previews along with extending in context video compilation. These combinations have proven to be highly lucrative and well suited for VAI in both creation and scale.
Example above demonstrates using video artificial intelligence to create and control audio and video transitions between clips. The video length and segments being used are determined based on video scoring generated via VAi.
Another exciting use case is making video libraries searchable. Finding specific people, actions, scenes within video opens the doors to advanced video discovery. A whole new world opens up with amazing possibilities of cross-reference and extracting information otherwise trapped in old school liner video metadata.
Solving the problem of creating net new video from existing video assets opens up a monster revenue source while extending your overall video lifetime value. Taking long-form video or live content and compressing to meaningful short-form mobile-ready is a game-changer. VAI creation does not require expanding editorial budgets or delaying critical production windows.
Organizing content to be contextually relevant will always be a key factor in ensuring meaningful content flow. What clip’s and scenes logically flows together are subjective unless you are directly measuring audience response. This is why utilizing an active feedback mechanism to improve relevance is important. Here we address some top challenges,
There isn’t a lack of challenges facing video creation at scale. For sure content specifically created for social platforms that harness the human connection and social dynamics will produce a higher quality product. However, how do you scale this type of creation? Combining AI with video provides the opportunity to develop a video learning system to assist in delivering quality output that builds on itself over time. Hence, why a learning system is ideal for video creation based on reproducible tasks.
Here we describe the strategy of video “fracking” your entire library. The ability to organize video libraries, extract cross-referenced videos, annotate video content and create relevant net new video makes MicroClips possible. Manipulating video at scale using artificial intelligence creates an enormous revenue opportunity. The ability to learn what works and adjust video assets requires thinking differently about how content is produced. The possibilities are endless over many content types, moreover, digital enables measuring levels of audience engagement like never before.
Above is a sports MicroClips compilation created using AI and is derived from many videos. MicroClips are created using incontext high value action sequences. These clips are places together to create the MicroClips compilation. What’s exciting is the variety of compilation that can be created and watching what the audience finds most entertaining. Highlights and compilations have garnered the highest play and share rates. Some have even “gone viral” !
Want to see more exciting Sports MicroClips in action?
The ultimate goal is to provide a cost-effective video creation process while improving viewability, quality and consumer play rates. Leveraging how consumers connect and the time they have available are additional driving factors to a smarter creation process.
There are many big bets being made as well as major networks investing in dedicated Snapchat teams to capture a slice of this multi-billion dollar pie and prove the model out.
Quality content and great storytelling require a human touch for now but for large-scale video organizing and creation, the future is bright for video artificial intelligence. The digital video industry is going through a major transformation and the consumers are the winners. We have seen the networks being put on notice with the incumbents not only producing top shows, receiving awards, drawing in compelling audience sizes but also leveraging advanced technology faster improving the consumer experience. The publishers that harness the competitive advantage of VAI creation will deliver quality video to market faster with higher relevance all resulting in greater consumer retention and revenue.