Research On Photon Mapping Clobalillumination Method

Digital animation and movie industry has rewarded a rapid development, and virtual reality technology has gradually come into our lives. In order to give people more realistic viewing experience, realistic rendering technology plays more important role in the production of animation movies and games. More realistic rendering effect makes more and more excellent film and animation works continue to emerge. The visual impact of global illumination is essential for photo-realistic rendering. Global illumination is a significant part of realistic rendering, and it has been focused by researchers for many years.However, rendering algorithms, especially global illumination algorithms, are computationally complex and time-consuming. The virtual scene used in the moves generally contains a large number of complex models, materials, lighting, etc, and the frame rate is required no less than 24 frames per second in order to ensure the continuity of animation. Therefore, fast and high quality global illumination has been the central goal of photo-realistic image synthesis for a long time.Global illumination computes not only the lighting directly from the light sources but also the lighting reflected by other objects in the scene and the lighting emitted by the surface itself. In computer graphics, the general rendering equation is used to solve the global illumination problem. In reality, we can see an object because lighting can be reflected into our eyes from the object. The rendering equation describes the relationship between the incoming light energy and the outgoing light energy on the surface of an object.Photon mapping is an extension of ray tracing method that makes it able to efficiently compute global illumination effects, such as caustics, ambient occlusion, color bleeding, soft shadows and soft indirect illumination in participating media. Photon mapping can be summarized in two main stages:photon map generating and radiance estimation. During the first stage, the photons are generated from the light source, emitted into the scene, scattered when intersecting with surfaces and stored in the photon maps at the diffuse intersection. In the second step, the photon maps are used to estimate the radiance at each shading point.Photon mapping algorithm has been developed over the past two decades as an active research field, which is an effective method for computing global illumination. The deficiency of precise details and efficient rendering are still the main challenges of photon mapping. We review recent works and classify them into a set of categories including radiance estimation, photon relaxation, photon tracing, progressive photon mapping, and parallel methods. The goals are giving readers an overall introduction to photon mapping and motivating further research to address the limitations of existing methods.Firstly, we focus on the bandwidth selection process in the second phase, as it can affect the final quality significantly. Poor results with noise arise if few photons are collected, while bias appears if a large number of photons are collected. In order to solve this issue, we propose an adaptive radiance estimation solution to obtain trade-offs between noise and bias by changing the number of neighboring photons and the shape of the collected area according to the radiance gradient. Our approach can be applied in both the direct and the indirect illumination computation. Finally, experimental results show that our approach can produce smoother quality while keeping the high frequency features perfectly compared with the original photon mapping algorithm.Secondly, we describe a modified progressive photon mapping method applied with sample elimination in order to pursue accurate results and accelerated iterations, which is called progressive photon elimination(PPE). An elimination status tree is proposed for PPE, which keeps the information using in eliminating and can be updated along with the multi-pass process. By using the status tree, our method can obtain a uniform photon distribution in each iteration by eliminating a certain amount of photons, and the tree is improved to adapt to the complicated photon distribution. This strategy also facilitates the parallel elimination with a priori elimination ratio in the status tree. In parallel block processing, equilibrium and edge issues are resolved. The experimental results show that our method can save about half of the number of iterations to achieve the same visual effect compared with PPM.Finally, we propose a parallel photon searching algorithm by using radiance estimation approach for coherent shading points on the Intel Many Integrated Core (MIC) Architecture. During photon searching step, each shading point needs to search the photon-tree to find k-neighbouring photons for reflected radiance estimation, which is time consuming. In order to efficiently use single instruction multiple data (SIMD) units, shading points are clustered by similarity first (every cluster contains 16 shading-points), and an initial neighbouring scope is searched from the photon-tree for each cluster. Then we use 16-wide SIMD units by performing k-NN searching from the initial neighbouring scope for those 16 shading-points in a cluster in parallel. We use the method to simulate some global illumination scenes on Intel Xeon processors and Intel Xeon Phi coprocessors. The comparison results with existing photon mapping techniques indicate that our method gives significant improvement in speed with the same accuracy.