利用光强的可叠加性，本文提出了将光场分解为不随场景变化的子光场的思想，使得物体、光源、观察点都可自由变动的动态场景绘制也能利用基于图象的绘制技术之优势。依靠漫反射与观察方向的无关性，本文还提出了漫反射光场的四维表示法，并且利用物体对不同点光源的漫反射之间以及环境光与漫反射光之间的内在联系，只用一个单位强度的单色点光源为物体构造漫反射光场，从而节省了存储空间和预处理时间。利用其数据结构的一致性，环境光场、漫反射光场和深度场中对应点的数据被合并成５维矢量以便更有效地压缩。本文给出的绘制算法，能准确地计算出物体

According to the fact that radiance may be superposed, the paper raise the idea of decomposing the light field into sub light field that do not change as scene changes, which extend the preponderance of light field rendering to dynamic scenes where coordinates and direction of object, light and view point in addition with the variation of light intensity can be modified freely. By exploiting the independence of diffuse light on viewing direction, the paper also proposes a four dimensional representation of diffuse light field. Considering the intrinsic relationship between diffuse reflections to different spot lights and between ambient light and diffuse light, all the diffuse light fields caused by different spot lights for an object can be recorded with a single diffuse light field generated by a pure color light with unit intensity, which decreases storage and preprocessing. Owing to their coherency of data structure, corresponding points of ambient light field, diffuse light field and depth field (used to compute shadow) are combined to 5 dimensional vectors that are more efficient to be compressed with vector quantization. The rendering algorithm given in this paper can accurately compute typical characteristics of dynamic scene as changes in surface color and shadow.