Unbiased Photon Gathering In Heterogeneous Media

Our world has gotten into the digital era now.The growing demand of market arises the amazing prosperity of animation film and video game industry,as well as the rapid development of photo realistic rendering technology.One of the most important and challenging topics in this field is about the efficient rendering technology for heterogeneous participating medium like smoke,cloud and flame.However,there is no satisfying solution for unbiased rendering technology in heterogeneous medium because of its complex optical properties.This thesis proposes a method for unbiased photon gathering in heterogeneous medium.It achieves a solid extension of original UPG by transforming the participating medium from heterogeneous to "virtually" homogeneous based on the idea of Woodcock tracking.We also combine UPG with bidirectional path tracing(BDPT)algorithm to compensate for the intrinsic drawbacks of UPG Specifically,we use the multiple importance sampling(MIS)technology and discuss the weight distribution strategy of it.Experiment result shows that our algorithm performs better than PM in terms of unbiasness and converges more quickly than BDPT.When the UPG+BDPT framework being used,the performance is further improved and the variance is obviously reduced.As a result,the advantages of UPG and BDPT are both fully exploited with our MIS weight strategy and more favorable rendering result can be achieved.