Scene depth reconstruction on the GPU: A post processing technique for layered fog

Abstract

Realism is a key goal of most VR applications. As graphics computing power increases, new techniques are being developed to simulate important aspects of the human visual system, increasing the sense of 'immersion' of a participant in a virtual environment. One aspect of the human visual system, depth cueing, requires accurate scene depth information in order to simulate. Yet many of the techniques for producing these effects require a trade-off between accuracy and performance, often resulting in specialized implementations that do not consider the need to integrate with other techniques or existing visualization systems. Our objective is to develop a new technique for generating depth based effects in real time as a post processing step performed on the GPU, and to provide a generic solution for integrating multiple depth dependent effects to enhance realism of the synthesized scenes. Using layered fog as an example, our new technique performs per pixel scene depth reconstruction accurately for the evaluation of fog integrals along line-of-sight. Requiring only the depth buffer from the rendering processing as input, our technique makes it easy to integrate into existing applications and uses the full power of the GPU to achieve real time frame rates. © Springer-Verlag Berlin Heidelberg 2007.