Computer graphics and computer vision have experienced remarkable advancements in 3D reconstruction and novel view synthesis, primarily propelled by the emergence of neural networks. In this survey paper, we provide a comprehensive overview of the state-of-the-art techniques in neural rendering for 3D reconstruction and view synthesis. We discuss datasets and metrics for evaluation, trace the evolution from classical methods to neural networks, explore advancements in image rendering, and focus on Neural Radiance Fields (NeRF). We cover NeRF fundamentals, efficiency, sparse data handling, dynamic scenes, composition, and application-specific NeRFs. We conclude by summarizing key findings and identifying future research directions.
- Sanidhya Singal
- Manas Sharma
- Ritika Kishore Kumar
- Krish Rewanth Sevuga Perumal