Two factors are required for a virtual reality system; realism and immersion. Since realism is determined by a realistic image, objects in a virtual environment should be represented with details by their own characteristics. Also, reflection on light must be considered. On the other hand, immersion is defined as realistic movement and fast interaction. Thus, virtual environment should be rendered as fast as possible. Regular frame rate should also be kept to make moves as intended by user.
In a complex virtual environment composed of object with various levels, the appropriate level for individual object should be assigned according to the view point, i.e., the selection of level-of-detail is needed. In this paper, the problem of determining the levels of detail is formalized in terms of error, cost and L-distance. Error is defined as the difference between the current level image and the highest level image, cost as rendering time, and L-distance as the level difference between previous and current level. To solve this problem, a well-distributed adaptive algorithm is proposed.