|dc.description.abstract||This paper examines the possibilities of improving fidelity of the rendered 3D model images integration into the given photograph.
This problem is most commonly tackled by designers, architects, and other professionals who have a need for visualization of virtual content. During the integration of CGI and photography, designers often try to avoid the issue of matching the shadows both of the photograph and the model which is being, because of the complexity of the restitution process, which demands additional data besides the photograph itself. This is why we often encounter solutions which skilfully manage to avoid dealing with this issue, by choosing the light (diffused, cloudy, twilight) that does not give clearly visible shadows, or that is set in such a way that frontal areas are lighted, while the shadows remains sheltered. If necessary, shadow matching is mostly done approximatively, so that any illogicality remains hidden from the eye of both non-skilled users and expert ones as well. Thus, the aim of this research is to improve the ways of the rendered image integration into photograph by applying the geometry knowledge, which enables matching the shadows more accurately, according to the requirements of the photograph. In the example of the photograph of the city square in Čačak (Serbia) together with a sculpture model made as a composition of geometric shapes, chosen to be placed on a given square in honour of geometry, a solution to this problem was given.
This research is developed as a continuation of study within the subject of Visualization and Presentation of 3D Models in Geodesy, included in the program of the 3rd year of studies at the Faculty of Civil Engineering in Belgrade. As one of the methodological units in lectures and at the practical exercises, the problem of restitution of object’s dimensions from the photograph is consider followed by the task of integrating the independent 3D model of an object (or objects) using its static 2D image – render.||sr