Gradient-domain Volumetric Photon Density Estimation

Adrien Gruson (1, 2) , Binh-Son Hua (2, 3) , Nicolas Vibert (4) , Derek Nowrouzezahrai (4) , Toshiya Hachisuka (2)

1 JFLI - Japanese French Laboratory for Informatics
2 UTokyo - The University of Tokyo
3 SUTD - Singapore University of Technology and Design
4 McGill University = Université McGill [Montréal, Canada]

Gradient-domain rendering can improve the convergence of surface-based light transport by exploiting smoothness in image space. Scenes with participating media exhibit similar smoothness and could potentially benefit from gradient-domain techniques. We introduce the first gradient-domain formulation of image synthesis with homogeneous participating media, including four novel and efficient gradient-domain volumetric density estimation algorithms. We show that naive extensions of gradient domain path-space and density estimation methods to volumetric media, while functional, can result in inefficient estimators. Focussing on point-, beam- and plane-based gradient-domain estimators, we introduce a novel shift mapping that eliminates redundancies in the naive formulations using spatial relaxation within the volume. We show that gradient-domain volumetric rendering improves convergence compared to primal domain state-of-the-art, across a suite of scenes. Our formulation and algorithms support progressive estimation and are easy to incorporate atop existing renderers.

Mots clés: gradient rendering; participating media

Domaines: Synthèse d'image et réalité virtuelle [cs.GR]

Adrien Gruson, Binh-Son Hua, Nicolas Vibert, Derek Nowrouzezahrai, Toshiya Hachisuka. Gradient-domain Volumetric Photon Density Estimation. ACM Transactions on Graphics, 2018, 37, ⟨10.1145/3197517.3201363⟩. ⟨hal-01796774⟩