Reproduce “Physics-Based Inverse Rendering using Combined Implicit and Explicit Geometries” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Guangyan Cai, Kai Yan, Zhao Dong, Ioannis Gkioulekas, Shuang Zhao
Researchers introduce a two-stage physics-based inverse rendering pipeline that integrates implicit and explicit geometric representations to reconstruct 3D objects and their spatially varying materials from images. This method effectively recovers complex object topologies and fine geometric and appearance details, even from simple initial shapes, while accurately accounting for advanced light transport effects.
Key insights
- Differentiable iso-surface extraction provides a crucial bridge, allowing physics-based differentiable rendering techniques designed for meshes to be effectively applied to topologically flexible implicit surfaces.
- A sequential, hybrid approach that leverages implicit representations for initial topological capture and explicit representations for subsequent detail refinement offers superior robustness and fidelity compared to methods relying solely on one representation type.
- The proposed method enables the reconstruction of complex objects under advanced light transport conditions, such as environmental illumination and soft shadows, which were challenging for prior implicit-based differentiable rendering techniques.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.