基于栅格的全局光子图重建

A Grid-based Reconstruction of Global Photon Maps

基于光子图的光子映射算法能产生高质量的照片级图像。对于光照复杂的场景,光子图需要存储大量光子以提高生成图像的质量,这不仅占用大量的内存空间,而且光照估计的时间长。论文提出基于栅格的全局光子图重建的算法,即在光子包围盒被栅格化后,其非空栅格中一定比例的光子被用来重建新的光子图,并保证重建前后栅格内光子能量和守恒,这使得重建前后光子图的光照估计的效果相近。通过增加特定栅格中的重建光子数目,能有效减少由几何偏差引起的光照估计误差,增强直接聚焦(焦散)和间接聚焦光照的绘制效果;并使用简单方法检测生成图像中少量噪声,增加少量采样即可有效减少相应的噪声。全局光子图重建算法的计算成本低,并保持生成图像的视觉独立性。

Photon mapping based on photon maps can efficiently render high-quality photorealistic images. It stores many photons in photon map to improve rendering qualities of complex illumination, leading to high memory overhead and long reconstruction time. This paper presents a grid-based reconstruction of global photon maps. We divide bounding box of photons into a number of grids. Based on the number of photons, we take a certain rate of photons in grids to reconstruct global photon map, and ensure conservation of total energy of photons in each grid before and after reconstruction, which make estimated illumination effects of photon maps before and after reconstruction almost same. By adding photons in special grids, we effectively reduce illumination error due to all kinds of geometry biases and enhance rendering effects of direct and indirect caustic. This paper checks a few noises in rendering pictures by simple method, and adds a few samples to effectively reduce these noises. Our reconstruction method of photon maps takes fewer time and ensures viewpoint independence of rendering pictures.