University of Richmond releases AI research tool covering more than 2,000 films
DataReel lets scholars explore patterns in editing, framing and audio. Understanding why those choices matter remains the research challenge.
Listen to this story
The audio brief
Story brief
3 key pointsDataReel gives researchers an open-access visual interface for comparing AI-generated measurements across more than 2,000 films. Released by the University of Richmond’s Distant Viewing Lab on November 20, 2026, it records features such as shot duration, framing, camera movement, and the share of speech or music. The dataset can support research into film form, but it does not establish how those choices affect viewers; those questions remain to be tested.
- 01
The tool also describes scenes, identifies onscreen characters, and counts people appearing in an image.
- 02
The project received support from Schmidt Sciences and includes collaboration with UC Berkeley and Bowdoin College.
- 03
Lab co-directors Lauren Tilton and Taylor Arnold bring digital humanities and data science expertise to the project.
Film scholars can now explore shot length, camera movement and sound across more than 2,000 films through an interactive visual tool. The University of Richmond’s Distant Viewing Lab released DataReel on November 20, 2026, offering open-access AI analysis intended to support research into how films are constructed and how their techniques vary.
From scenes to measurable patterns
DataReel uses AI models to capture film data automatically, then presents that information in an interactive, visual format. According to the university, the analysis covers both what appears onscreen and formal choices such as framing, editing and the use of sound.
- Scenes and people: The tool captures scene descriptions, identifies characters onscreen and measures how many people appear in the image.
- Shots and movement: It measures shot length, how much the camera moves and framing distance—the distinction between a close-up and a long shot.
- Speech and music: Its audio measurements include how much of a film consists of speech or music, adding sound to the visual record.
The university points to several questions those measurements could help scholars explore: how close-ups influence psychological engagement, how pacing affects narrative tension, and how visual and audio techniques differ across genres. Those are proposed research uses, rather than demonstrated answers about how audiences respond.
Current AI models can analyze two-dimensional images, but not how or why they were filmed in a particular way,
Lauren Tilton, professor of digital humanities and co-director of the Distant Viewing Lab
Beyond recognizing objects
Tilton describes the lab’s research ambition as moving beyond object recognition toward analyzing patterns in moving images. She frames that work as relevant to scholarly communities studying digital media, not just film researchers.
The lab brings together two disciplines through its directors: Taylor Arnold is a professor of data science and statistics, while Tilton is a professor of digital humanities and directs Richmond’s Center for Liberal Arts and AI.
Schmidt Sciences awarded support for the project, which includes collaboration with UC Berkeley and Bowdoin College. The released tool gives scholars an open-access way to examine the resulting film data; the questions about engagement, tension and genre remain subjects for research.
Sources
- news.richmond.eduUniversity of Richmond Distant Viewing Lab Releases DataReel, a New AI Tool for Film Research
Reader comments
Newest comments first. Replies stay oldest first.