Video editing can be an arduous process, even more so if you’re editing large amounts of footage for an event or wedding. A Disney research team is working on an inventive video technology that combines multiple video feeds in a way that both makes sense and obeys the basic rules of cinematography.
The program follows an algorithm that selects footage based on what it understands to be the most interesting content in the scene. For example, if the event were a basketball game, the algorithm would select footage based on which camera has the best shot of the ball at the time, while also staying within the conventions and rules of cinematography, such as not breaking the 180-degree rule and avoiding jump cuts. The algorithm also maintains a short shot-duration, which allows for a more cohesive edit that flows naturally and doesn’t take the audience out of the video.
The computation needed to achieve these results takes several hours. Contrast this with professional editors, who, using the same video feeds, took on average more than 20 hours to create just a few minutes of video. But fear not dear editors, this technology cannot make up for the narrative or technical complexity that a human editor can achieve. Think of it as a convenient assistant to editors tasked with large amounts of footage.
And there is more.
Imagine seeing an entire music concert via footage compiled by the members of its crowd. This method can take all of those disparate YouTube recordings generated from the crowd and stitch them together to maintain the best vantage point throughout the entire concert.
Although there are already other apps and methods for automatically or semi-automatically combining footage from multiple camera feeds, they do not have the specificity of Disney’s. Other methods either fail to follow the action or they cannot take cinematographic guidelines into consideration because they can’t map the orientation of each camera.
Now don’t go on assuming this will instantly turn you into a Martin Scorsese.