Methods and algorithms of computational mathematics and their applications

Adaptation of Global Illumination Computation Methods Based on the Radiosity Algorithm to the Frame Graph Architecture

A. S. Shcherbakov

Lomonosov Moscow State University, Faculty of Computational Mathematics and Cybernetics

Abstract: This paper examines the integration of the radiosity method into the frame graph architecture to optimize GPU memory consumption. Methods for adapting the multi-bounce matrix method, the local matrix method, and the temporal radiosity method are proposed, considering resource management requirements on the GPU. A scheme for reusing temporal memory across different stages of global illumination processing is described, which helps reduce video memory load and minimize overhead during algorithm execution. An experimental study on test scenes confirms the effectiveness of the proposed approach and demonstrates a reduction in memory consumption.

Keywords: global illumination, computational graphs, computer graphics.

UDC: 004.925.3

Received: 02.03.2025
Accepted: 03.03.2025

DOI: 10.26089/NumMet.v26r207