We represent a complex volumetric cloud using traditional grid-based methods (left and right, 1024³ voxel grid resolution, 4GBs) and our primitives-based representation using Gaussian kernels (middle, 5.6k primitives, 224KBs), and render it with volumetric path tracing. Our method achieves substantial speedups thanks to the analytical transmission estimation and sampling, our efficient rendering approach and its extremely compact representation. When compared to the original asset, at a potential cost of detail (Figure 7), we provide large performance and memory compression benefits. Asset is part of the Walt Disney Animation Studios cloud dataset (CC-BY-SA 3.0). Rendering times reported on a NVIDIA A6000.

We propose a compact and efficient alternative to existing volumetric representations for rendering such as voxel grids. Inspired by recent methods for scene reconstruction that leverage mixtures of 3D Gaussians to model radiance fields, we formalize and generalize the modeling of scattering and emissive media using mixtures of simple kernel-based volumetric primitives. We introduce closed-form solutions for transmittance and free-flight distance sampling for different kernels, and propose several optimizations to use our method efficiently within any off-the-shelf volumetric path tracer. We demonstrate our method in both forward and inverse rendering of complex scattering media. Furthermore, we adapt and showcase our method in radiance field optimization and rendering, providing additional flexibility compared to current state of the art given its ray-tracing formulation. We also introduce the Epanechnikov kernel and demonstrate its potential as an efficient alternative to the traditionally-used Gaussian kernel in scene reconstruction tasks. The versatility and physically-based nature of our approach allows us to go beyond radiance fields and bring to kernel-based modeling and rendering any path-tracing enabled functionality such as scattering, relighting and complex camera models.

Mixture Models of Arbitrary Kernels: Epanechnikov Volumes are Faster and Better Fits for Hard Surfaces



Radiance Field Optimization and Rendering



Inverse and Forward Path-traced Gaussian Volumes



Analytic Transmittance Estimation



Level of Detail and Performance-Quality Control

Downloads



Acknowledgements



Collaborations