全局光照环境中的逆向绘制

Inverse Rendering in a Global Illuminated Environment

提出一种恢复场景中所有物体的反射特性的方法.算法以一个全景图、场景的完整几何模型和光源信息为输入,结果是场景的一个完整反射模型.恢复是以逐步求精的方式进行的.先假定物体表面是漫反射面并生成一幅全景图,然后算法逐步迭代,比较绘制的全景图和原始全景图,如果有的物体的误差超过某个阈值,算法为 他们选择更负责的反射模型.最后,场景中的每个物体有了一个合适的反射模型,可以在新的光照和视点条件下绘制,旧的物体可以从场景中移去,新的物体可以添加到场景中.对漫反射、各向同性反射和各向异性反射纹理面的纹理恢复也作了深入研究;高光和阴影的影响也能基本去除.

A method is proposed to recover the reflectance properties of all objects in an environment. The input of the algorithm is a 3D geometric model of the scene, a panorama of it, and the information of light sources. The result is a full model of the reflectance properties of the scene. Recovery is done in a progressive refinement manner. At the very beginning, supposes all surfaces are diffuse and generates a new panorama. Then the new rendered panorama and the original one are compared iteratively. If the differences of some objects are greater than a threshold, more complex reflectance models are chosen for them. Finally, each object in the scene has a proper reflectance model. The scene can be rendered under novel lighting or viewing conditions, old objects can be removed from the scene, and new objects can be augmented into the scene. Special efforts have been done on recovering textures of textured diffuse, isotropic, and anisotropic surfaces; shadows and highlights are eliminated almost clearly.