From the campfire and cave wall, to the cinema and the tiny screens that dominate our lives — telling stories is a big part of what it means to be human. Along the way, technology has continually expanded the canvas of storytelling, adding new layers to the mix — like sound, moving pictures, and color. One layer that hasn’t been captured yet is that layer of what we’re thinking and feeling while watching, or even making, videos. There are characters, relationships, motivations, character arcs, intentions, emotions… What if that whole story layer could be captured? That’s why we started OZU — to figure out how to record that story layer inside videos — and help people find and tell their own stories.
In one of my earliest jobs in film, I worked with the filmmaker Michal Goldman, shooting footage for her documentary about a new tuition-free private school in Dorchester, Massachusetts. The story Michal wanted to tell was about adversity — about families trying to give their kids a better shot, about a school trying to serve the forgotten — but capturing that story was challenging. We’d spend 6–12 hours a day in the school and in the community, and shot over 170 hours during the school’s first year. After each day of shooting, we’d turn the camera on each other and talk about what we’d seen — what we’d learned — what we’d felt. But what we couldn’t do was search through what we’d shot — organize it while we were shooting — or have a conversation with it to see how deep our stories went. It was so simple in my head. We knew that every Tuesday at 9am was math class — and we knew math class was full of stories. There were shots of students banging their desks and tearing up papers in frustration, chewing on pencils and tapping feet deep in thought, and grinning ear-to-ear with pride. But we couldn’t just google through our footage, or chat with it to find those scenes. Everything needed to be watched, logged, and organized. It wasn’t something we could think aloud with.
The problem was that the story layer inside of the video we were shooting was locked up. It needed to be mined by hand, as if the ‘math class’ scenes had to be extracted like some rare mineral across 170 hours of footage.
Video is made up of more than image and sound. It’s about the people, their intention, the look in someone’s eye, or the words you wish the character would say that they never say.
Think about the videos on your camera roll. Or think about your favorite show. There’s a world of stories within the video in our lives — yet there’s no way to see that story layer — or make it easy to search through. It’s not on IMDB — or on Reddit. It’s somewhere in our heads and hearts.
What if this story layer could be recorded?
What if all the videos ever shot had this story layer captured and searchable?
That’s the world we want to live in at OZU.
To that end, OZU has invented the Precorder — a video AI that captures the story layer from video content. We’ve built the Precorder after researching and developing a range of Machine Learning models designed to understand film, television, and long-form narrative video. The Precorder captures:
- Art direction, including cinematography, lighting, and color
- Characters
- Locations
- Visual context — objects, props, setting, backgrounds
- Sounds — dialogue, music, foley
How do we do this? The Precorder uses OZU’s custom models along with open source models. We use a combination of open source and customized domain expert models trained on datasets we’ve built with expert storytellers. We’ve also designed an interface and programming language for people to explore and create using Precordings. (More to come on both of these things soon!)
Why are we doing this? We believe media evolves with technology- and just as sound was added to moving pictures, and then color — Precording adds a story layer to video — which will open up new creative storytelling possibilities.
Who are we? OZU’s founders are Anton Marini, Rahul Somani, and Eli Chapman. We are a small multi-skilled team of cinema lovers with a passion for building technology products. We started OZU to make it easier for people to read and write with moving pictures — to use the grammar of cinema and think aloud with video. Thanks for joining us on our journey. We’re excited to help you find your story.
Sign up for our friends and family launch here
