Martin Jakab and László Szécsi
Supervisor(s): Dr. Szécsi László
Budapest University of Technology and Economics
Abstract: This paper presents a novel approach to volumetric shadow mapping aimed at reducing the computational redundancy inherent in conventional methods. Traditional volumetric shadow maps evaluate the same depth samples along all primary rays that coincide in the depth map space. Our method pre-processes the shadow map and gathers discrete shadow segments - key intervals that capture the essential shadow-casting characteristics of objects - along possible primary ray images in the depth map. This requires accurately determining the angular and radial ranges from the light source's perspective, ensuring that only the relevant portions of the scene are processed during shadow calculation. A major benefit of this approach is that the computational cost scales with the scene's depth complexity rather than the number of ray casting samples required for a given quality level. Preliminary evaluations indicate that this approach not only streamlines the shadow computation process but also maintains high visual fidelity.
Keywords: Real-time Graphics, Rendering
Full text: Year: 2025
