Abstract: A new shading method is presented to avoid complicated modeling processes and heavy calculations. Based on the analogy of source images and destination images, the shading knowledge of source images under all kinds of light conditions is abstracted, and then applied to the destination images. On condition that the shaded objects in source images and destination images are perfect diffuse objects and similar, the rendering results are reasonable. This method does not need a lot of original data such as details of geometric model or depth of images. It can work quickly and control the shading effects flexibly, is suitable for acquiring shading effect on discrete viewpoints with variable lighting conditions. It has been applied to the Digital Rebuilding and Virtual Navigation System of Dunhuang Culture Relics with good results.