Automated Software Testing Technologies for Realistic Computer Graphics

Abstract This paper discusses automated software testing technologies for industrial systems of computer graphics and optical modeling. Test automation becomes vital in the case of limited resources and frequent releases of program versions, which is a typical situation for software developers. In this paper, we consider both regression testing methods for computing kernels of software complexes and testing methods for graphical user interfaces. The regression testing uses the Python scripting engine. Some methods for its parallelization are described, which can significantly reduce testing time. With stochastic methods being widely employed for lighting simulation, their computational results can differ, which complicates regression testing. In that case, we apply a certain threshold to compare the results. Automated tests for graphical user interfaces are based on the AutoIt tool. In addition, we consider some approaches to testing graphical user interfaces of systems implemented as plug-ins for closed-source CAD complexes, the source code of which is not available to the developers of the automated tests.