Broadening View Synthesis of Dynamic Scenes from Constrained Monocular Videos
This addresses the challenge of generating stable and realistic renderings from constrained monocular videos for dynamic scenes, which is incremental as it builds on existing dynamic NeRF frameworks.
The paper tackles the problem of unstable novel view synthesis in dynamic Neural Radiance Fields (NeRF) under significant viewpoint deviations by introducing ExpanDyNeRF, which leverages Gaussian splatting priors and pseudo-ground-truth generation to enable realistic synthesis under large-angle rotations, significantly outperforming existing methods in rendering fidelity.
In dynamic Neural Radiance Fields (NeRF) systems, state-of-the-art novel view synthesis methods often fail under significant viewpoint deviations, producing unstable and unrealistic renderings. To address this, we introduce Expanded Dynamic NeRF (ExpanDyNeRF), a monocular NeRF framework that leverages Gaussian splatting priors and a pseudo-ground-truth generation strategy to enable realistic synthesis under large-angle rotations. ExpanDyNeRF optimizes density and color features to improve scene reconstruction from challenging perspectives. We also present the Synthetic Dynamic Multiview (SynDM) dataset, the first synthetic multiview dataset for dynamic scenes with explicit side-view supervision-created using a custom GTA V-based rendering pipeline. Quantitative and qualitative results on SynDM and real-world datasets demonstrate that ExpanDyNeRF significantly outperforms existing dynamic NeRF methods in rendering fidelity under extreme viewpoint shifts. Further details are provided in the supplementary materials.