Research On Stochastic Progressive Photon Mapping Based On Multiple Sampling

Since the development of computer graphics,graphics rendering has been a field chosen and continuously studied by many scholars.From the ray casting technology proposed by Arthur Appel to the stochastic progressive photon mapping technology proposed by Hachisuka et al.,the efficiency and robustness of rendering algorithms have been improving.In the decades of the development of graphics,graphics rendering technology has been widely used in the fields of pictures,videos,games and animation.The main purpose of graphics rendering is to render some virtual scenes for realistic images.Photon mapping algorithm is also widely used as a rendering algorithm that can render various realistic effects such as caustics,environment mask,volume scattering,etc.In order to improve the rendering efficiency and the authenticity of the rendering scene,and solve the problem of image noise in the rendering process,a multiple sampling technology based on stochastic progressive photon mapping is proposed.The main algorithm of graphics rendering is improved,and photon resampling technology is proposed.At the same time,photon resampling technology combined with our error method can efficiently render complex scenes and obtain higher quality images.According to the detail convergence problem existing in the stochastic progressive photon mapping algorithm,we improve the method.The improvement part includes the following aspects.First,photon resampling technology is proposed,which changes the search radius of hit points in sub-blocks many times.This improvement is conducive to improving photon coverage rate.Secondly,the concept of image space is used to divide images into uniform sub-blocks and extract the image features sequentially such that they show consistency and continuity.Thirdly,the distance error method and photon number error method are proposed.The error method is combined with photon resampling technology to select the appropriate sampling radius for the hit point and resample the photons within the search radius of the hit point for many times.Our improvement of the algorithm can further solve the convergence problem caused by the large difference between light and shade of photon distribution,reduce noise and improve the performance of the algorithm.In addition,the algorithm proposed in this paper can also be applied to other photon mapping rendering algorithms similar to the stochastic progressive photon mapping algorithm.Our works improve the usability and robustness of the algorithm.The multiple sampling algorithm proposed in this paper optimizes the details in the rendering process and simulates the approximate results of real illumination in the scene,thus generating more real images in the illumination scene.Experiments show that the multiple sampling algorithm has achieved obvious improvement in rendering efficiency and quality.