We have been developing an efficient and effective image generation system for "Mixed- Reality Traffic Experiment Space", an enhanced traffic/driving simulation system as one part of the "Sustainable ITS Project" at The University of Tokyo. Unlike conventional driving simulators, our system represents the view through image/geometry-based hybrid method to realize high photo-reality with less human cost. Images for datasets are captured from real world, and the view for the simulation system is created by synthesizing the image dataset and conventional polygon-based simulator. We are planning to apply this system to a sag zone in Tomei Expressway in Japan, where subtle changes of road gradient often causes traffic congestions. Using our system, we can re-create the situation with rich photo-reality and also expected to help human/psychological analysis of a driver. This paper describes the outline of our plan mainly targeting some technical problems and representing our proposing rendering method.