Adaptation of supersampling technology to increase the performance of real-time ray tracing image rendering

Abstract

The article presents an option for adapting the supersampling technology, or oversampling smoothing, to increase the performance of image rendering using real-time ray tracing. A modified approach with different resolutions is proposed. Instead of following the rigid quad tree grid structure inherent in textures and stored pixels, a triangular grid is used to collect samples. This triangular mesh is actually superimposed on the traditional texture format used to collect the samples in the offset pattern and then returned to the traditional mesh for display on the screen. We believe that using this triangular sample will require fewer samples, improving the efficiency of the algorithm without compromising the quality of the final rendering. The results of testing the adapted algorithm relative to the unmodified rendering method are presented.