 |
Abstract: NeRFs and Gaussian Splatting for Sets at Virtual Studio [DE]
Christopher Grimm, NeRFs and Gaussian Splatting for Sets at Virtual Studio, Hochschule Düsseldorf, Bachelor thesis, 07.11.2024.
The creation of 3D assets is one of the most time consuming aspects of production in a virtual
studio, in addition to being costly. Neural Radiance Fields (NeRF) and Gaussian Splatting
(GS) offer the ability, to capture real scenes and convert them into 3D models, tough by using
alternative methods to traditional polygon models. This way, storage Space, performance
and time can be saved with simple tools, as well as a real location that shall be used a set in
a virtual studio, with little work and little to no costs and a high quality result. The Aim of
this thesis is to find out, how these models can be used in a virtual studio, as well as how the
best possible quality can be achieved for them.
Keywords:
Neural Radiance Field, NeRF, Gaussian Splatting, GS, virtual studio, set, machine learning, AI
Supervisor:
Prof. Jens Herder, Dr. Eng./Univ. of Tsukuba
Prof. Dr.-Ing. Thomas Bonse
Location:
The research took place at the Virtual Sets and Virtual Environments Laboratory.
|
|
