An effort to develop a Digital Elevation Model (DEM) for small scale spatial objects is presented in this paper. The motivation of the research is to visualize a small spatial object as detail as possible in 3D, thus the presentation of the object in term of a computer model is similar to its presentation in reality. To reach this objective, the research was conducted in four stages i.e. elevation data retrieval, 3D visualization and its enhancement, and the development of blank data removal. Elevation data is obtained from SRTM dataset that has 3 arc-second or approximately 90 meters data resolution. Data obtained from SRTM is then visualized in 3D, in which visualization is enriched with view angle setting. Results of visualization show that enhancement to SRTM data set is required to present elevation data in 3D. It is due to the existence of blank data contained in the elevation data retrieved from SRTM. Therefore, this paper proposes two methods to enhance SRTM dataset i.e. population average and neighbors average, in order to counter the existence of blank data. The first method uses the population average of statistic to eliminate blank data, while the latter relies on the average value of its neighboring pixels. Comparison among those methods is held in this research to disclose the performance of each method. Result of comparison shows that the method based on neighbors average outperform population average method to eliminate blank data contained in SRTM data set. Thus, neighbor average delivers better 3D visualization for small scale spatial object. Keywords: Digital Elevation Model, 3D Visualization, Spatial