An industrial implementation of modern inpainting techniques for depth-based 3D film editing

Research project in the area of Image and Video Analysis & Synthesis.

Keywords: Inpainting, Novel Views, 3D film.

About this Project

The goal of this project is the implementation of a software prototype which enables the adaption of depth perception in 3D films in unprecedented quality. In this project, Emotion3D and Vienna University of Technology are planning to solve the inpainting problem by a highly innovative approach. The problem typically arises in post-production, when geometric distortion occurs in the process of depth perception adaption. Through this distortion, areas are being created in which image data is missing and these blank areas need to be filled through intelligent copying of neighbouring image areas (inpainting). The requirements are high, since visual errors are easily noticeable and not acceptable for high-quality film production. Despite the fact that there are technically advanced inpainting techniques available, their application in complex areas like 3D film has been limited due to the immense computing effort and low robustness. The core of this project is a research project by Princeton University presenting a new method for the structural editing of images, which represents a significant step forward in solving this problem efficiently and adequately.

Project Partners

Emotion3D GesmbH (Wien)

Funding provided by

Zentrum für Innovation und Technologie (ZIT)

Additional Information

Received funding from the Technology Agency of the City of Vienna (ZIT)