It is impossible to visualize large data on portable devices like smart phone. To make it possible, data reduction is required and the data size has to be reduced under a specified volume regardless of the original size. In this condition, the reduction methods based on the rendering resolution are better than the methods that have dependency of the complexity of the original shape. I have proposed a new reduction method based on an observer’s view resolution. The theory is to extracts only visible points from the viewpoint positions which are capable of observer sees. In this paper, I describe the technology, the prototype implement and the benchmark results.