Evaluating NeRF Fidelity using Virtual Environments

Neural Radiance Fields (NeRF) are a promising form of 3D Reconstruction that utilize sparse 2D imagery to recreate synthetic and real scenes. Since NeRF was first developed, numerous methods and techniques have worked upon the original algorithm to improve its accuracy, fidelity, and speed. We aim to evaluate the fidelity of a reconstructed scene, as well as evaluate the quality of the reconstruction.