Abstract. Nowadays, in light of the latest development in three-dimensional (3D) modeling technology, an essential role is given to the research and development of fully-automated or semi-automated processes in order to increase workflow effectiveness. A key challenge is thus to automate the process leading to the geometric model which supports the Building Information Modeling (BIM) or 3D-Geographical Information Systems (3D-GIS). This 3D model usually originates from image-based or range-based point clouds. This research is the beginning of the development of a 3D modeling approach that is semi-automatic, and possibly fully-automatic, by combining polygon surface fitting (polyfit) technique and monoscopic multi-image measurement system. With the advent of dense matching and Structure from Motion methods (SfM), point clouds can be generated from multiple images obtained from digital cameras. Then, to reduce the data and to allow for efficient processing, it is necessary to extract polygonal surface data from point clouds delivered by the dense matching process. The polygonal surface is then used for the basis of further manual monoscopic measurements which are achieved separately on each image to obtain more detailed 3D model. Next, this approach analyzed the polygonal surface deformations in comparison to the initial point cloud data. It can be seen how the resolution and noise of the original point clouds affect the subsequent Polyfit-based modeling and monoscopic measurements. The deformations and the accuracy evaluation have been undertaken using different open source software. Also, the geometric error in the polyfit-derived polyhedral reconstruction propagating to the subsequent monoscopic-derived measurements was evaluated. Finally, our modeling approach shows that it can improve the processing speed and level of detail of the 3D models achieved using existing monoscopic measurements. Typically geometric accuracy itself doesn’t have enough information to make accurate geometry model.