Techniques for increased realism in visual simulation and scientific visualization

Computer generated images representing real world data and simulations of real world events can be used to gain new insights into the physical world. This dissertation describes techniques which improve the quality of computer generated images of these types of data. One area of computer graphics which is investigated is real time visual simulation, the goal of which is to present the most detailed possible image of a scene while maintaining the screen update rate required to provide smooth animation of an image. This dissertation presents a method for rendering images in an air/sea flight simulation which provides additional visual cues to the viewer by providing an accurate, animated representation of the wake of a ship. Scientific visualization is another area of computer graphics which is investigated here, specifically the methods used to determine surface orientations (normal vectors) for use in shading calculations and surface patch generation. In this dissertation, commonly used techniques for determination of surface orientation of sampled data are evaluated, and new techniques are developed which are shown to provide more accurate rendering of fine detail present in the data. These techniques are applied both to the rendering of two-dimensional data as a three-dimensional height field, and to the rendering of images derived from three-dimensional medical Magnetic Resonance Imagery (MRI).