Panorama

Team

Dustin Escoffery, Wai Son Wong, Alexei Othenin-Girard, Wesley Souza, Devon Wyland

Advisor

Arnav Jhala

Research Labs

Computational Cinematics

NEW: Download Alpha version

Overview

Panorama is a game designed to collect images with well-defined visual features. Pictures taken through gameplay are automatically annotated for composition quality, and contribute to a corpus of domain-specific virtual photographs. Data from the Panorama Corpus is used to learn photographic preferences. We describe the design decisions in the development of Panorama, visual composition features modeled in the game, and results from an experiment to collect relative pairwise ratings on the images to learn how different features contribute to an individual's personal preference for photographic composition.

Video