Tuesday 12pm, 12 September 2017

Cs184 dk


Alex Hall

PhD Student - UC Berkeley


Over the last century, filmmakers have created a well-defined grammar of cinematography. Texts such as Daniel Arijon's 'The Grammar of the Film Language' lay this grammar out for film students, practitioners, & researchers. However, given the enormous corpus of films, it is difficult for a human to study a significant portion of extant films at the level of this grammar. We present a system that decomposes films into a semantically meaningful vocabulary. We then use this system to perform initial analyses of a 600 film dataset. We show that our system can recreate analysis by researchers such as James Cutting and scale those works up to any number of films with little to no additional human input. We also present future ideas for analytic methods using our pipeline, including prototype machine learning implementations for film classification and comparison tasks.


Alex Hall is a National Science Foundation Graduate Research Fellow & Graduate Student Researcher at UC-Berkeley working with Maneesh Agrawala (Stanford) and Alyosha Efros (UCB). His first undergraduate degree was in Theatre Performance from the University of Wisconsin-Whitewater with a minor in Strength & Conditioning. Years later, he earned his second undergraduate degree, also from UW-Whitewater, and was one of the first four graduates of the UW-Whitewater Computer Science Program. He earned a Certificate of Computer Science from the University of Wisconsin-Madison. A former professional Shakespearean actor, Alex brings an artistic perspective to Computer Science along with a love of story telling and the performing arts. His research is focused on building systems to help artists and art researchers explore and understand their craft. He also rescues cats.