Subspace-based Bidirectional Path Tracing Rendering

Global illumination rendering is one of the most important research areas in computer graphics,and ray tracing is the most representative algorithm in global illumination technologies.Ray tracing algorithms implemented have been widely used in the industry,but they are still confused by noise and perform poorly in some lighting scenarios.Bidirectional path tracing traces sub-paths from both the camera and the light source and connect them one by one to obtain a full light path for rendering,which can robustly handle various complex scenes.The probabilistic connection method is an extension for bidirectional path tracing and accelerates the rendering by selecting appropriate light sub-paths for connection.However,existing probabilistic connection algorithms require frequent distribution reconstruction and have an expensive overhead.This paper proposes a novel approach for constructing the light selection distribution in sub-path space to achieve probabilistic connections.We divide the sub-path space into multiple subspaces and classify the similar sub-paths into the same subspace,then the light sub-paths can be selected by a subspace-based two-stage sampling method,i.e.,first sampling the light subspace and then resampling the light sub-paths within this subspace.Our subspace-based distribution is free of reconstruction and provides efficient light selection at a very low cost.We also propose a method that considers the Multiple Importance Sampling(MIS)term in the light selection and thus obtains an MIS-aware distribution that can minimize the upper bound of variance of the combined estimator.Prior methods typically omit this MIS weights term.To address the problem that the connection strategy can’t sample the specularinvolved difficult path,we also propose a new sampling method called proxy sampling,which allows us to substitute vertices on a path at will.By using proxy sampling,we can retrace the uncertain vertices in the path connection,we perform good sampling for the specular-involved difficult path.We evaluate our algorithm using various benchmarks,and the results show that our approach has superior performance and can significantly reduce noise compared with the state-of-the-art method.