Real-time depth-of-field rendering using single-layer composition

In this paper, we propose a single-layer post-processing method for real-time depth-of-field rendering that uses single-layer composition. In the proposed method, blurring is achieved by gathering background pixels and scattering foreground pixels. Major artifacts in post-filtering techniques such as intensity leakage and blurring discontinuity are reduced by using two different blurring functions and the controllable parameter in the gathering process. The method can be entirely implemented in GPU parallelization to achieve the real-time performance required for virtual reality. The results of comparisons of our method with recent post-processing methods in terms of rendering quality and rendering performance indicate that our method generates realistic natural images and is also the fastest in terms of frames per second.