Feel3D: Free Viewpoint Video Making Mechanism for Capturing Premeditated Scenarios Using Pre-Built Models

Abstract

Free viewpoint video(FVV) recording is one of the emerging home entertainment mediums. When watching a free viewpoint video viewers are given the freedom to choose the viewpoint that they want to watch the video. Current FVV making methods depends on real-time video footage to synthesise new views or apply texture on models in the video. This approach faces challenges in producing structurally and textually correct views when video recording conditions change. This project focuses on using pre-built 3D models for FVV making. As opposed to conventional FVV making methods, proposed approach will produce better quality 3D FVV with less processing power requirements. In addition to that, the project will introduce a new framework to follow when making 3D free viewpoint videos. The 3D FVVs made with the proposed approach will allow e cient storage and bandwidth usage during transmission. Following sections will present about employed motion tracking method, use of 3D models and how all the parts are assembled together to produce the free viewpoint video. The proposed method is best suited for recoding a pre-planned scenario that happens in indoor environments. The latter stages of the report will discuss strengths and limitations as well as the e ect of this approach on commercial and research interests related to FVVs.