Fourier Splatting: Generalized Fourier encoded primitives for scalable radiance fields
This provides a versatile solution for bandwidth-constrained high-fidelity rendering in novel view synthesis applications.
The paper tackles the problem of quality downscaling in radiance field rendering by proposing Fourier Splatting, which uses Fourier-encoded planar surfels to enable scalable rendering detail through coefficient truncation, achieving state-of-the-art quality among planar-primitive frameworks with comparable perceptual metrics to leading volumetric methods.
Novel view synthesis has recently been revolutionized by 3D Gaussian Splatting (3DGS), which enables real-time rendering through explicit primitive rasterization. However, existing methods tie visual fidelity strictly to the number of primitives: quality downscaling is achieved only through pruning primitives. We propose the first inherently scalable primitive for radiance field rendering. Fourier Splatting employs scalable primitives with arbitrary closed shapes obtained by parameterizing planar surfels with Fourier encoded descriptors. This formulation allows a single trained model to be rendered at varying levels of detail simply by truncating Fourier coefficients at runtime. To facilitate stable optimization, we employ a straight-through estimator for gradient extension beyond the primitive boundary, and introduce HYDRA, a densification strategy that decomposes complex primitives into simpler constituents within the MCMC framework. Our method achieves state-of-the-art rendering quality among planar-primitive frameworks and comparable perceptual metrics compared to leading volumetric representations on standard benchmarks, providing a versatile solution for bandwidth-constrained high-fidelity rendering.