Evaluation of rendering algorithms for augmented reality using psychophysics and iCAM

• Authors: Timothy J. Hattenberger;Garrett M. Johnson;Mark D. Fairchild
• Venue: APGV '05 Proceedings of the 2nd symposium on Applied perception in graphics and visualization
• Year: 2005

Physically based global illumination rendering algorithms have matured to the point of being able to create physically and visually realistic imagery [Ferwerda 2003]. However, this realism comes at the expense of computation time, which is one reason these rigorous solutions have not been fully embraced, particularly by the entertainment industry. Previous visual experiments judging realism [McNamara 2000] have compared real and synthetic images created using different algorithms or varying the parameters within an algorithm. However, it is not necessarily valid to extend these results to different scenes or applications. The thrust of this research is to compare the visual differences and relative computation times between variations of several ray-traced based global illumination algorithms as applied to the problem of rendering synthetic objects into real imagery. Ultimately, this work will provide specific tradeoff information for the rendering algorithms used in this scene, a general pipeline for evaluating new algorithms based on specific criteria (e.g.visual realism), information regarding artifacts specific to realism in augmented reality, and the potential use of a computational vision model such as iCAM [Fairchild and Johnson 2004], in the loop. This poster presents the steps used to create the imagery and initial results.