Differentiable Ray

Differentiable ray tracing leverages the power of automatic differentiation to optimize rendering processes and related inverse problems across diverse fields. Current research focuses on applying this technique to improve 3D reconstruction (e.g., using neural implicit representations and Gaussian splatting), high-fidelity image synthesis (including novel view synthesis and relighting), and inverse problems in areas like radar and photoacoustic imaging. This approach enables faster and more accurate solutions compared to traditional methods, impacting fields ranging from computer graphics and medical imaging to 6G communication and autonomous driving through improved simulation and data analysis.