Zhang, YanciPajarola, RenatoMario Botsch and Baoquan Chen and Mark Pauly and Matthias Zwicker2014-01-292014-01-2920063-905673-32-01811-7813https://doi.org/10.2312/SPBG/SPBG06/037-048Hardware accelerated point-based rendering (PBR) algorithms have suffered in the past from multiple rendering passes; possibly a performance limiting factor. Two passes over the point geometry have been necessary because a first visibility-splatting pass has been necessary for conservative e-z-buffer visibility culling in the following point-interpolation rendering pass. This separation into visibility-splatting and point-blending, hence processing the point geometry twice, is a fundamental drawback of current GPU-based PBR algorithms. In this paper we introduce a new framework for GPU accelerated PBR algorithm whose basic idea is deferred blending. In contrast to prior algorithms, we formulate the smooth point interpolation problem as an image compositing post-processing task. This is achieved by separating the input point data in a pre-process into not self-overlapping minimal independent groups of points. As an extension of this concept, we can for the first time render transparent point surfaces as well on the GPU. For simple transparency effects, our novel algorithm only needs a single geometry rendering pass. For high-quality transparent image synthesis an extra rendering pass is sufficient. Furthermore, per-fragment reflective and refractive multilayer effects are supported in our algorithm.Categories and Subject Descriptors (according to ACM CCS): I.3 [Computer Graphics]: I.3.3 [Picture/Image Generation]: Display algorithms I.3.5 [Computational Geometry and Object Modeling]: Surface representations I.3.7 [Three-Dimensional Graphics and Realism]: Color, shading, shadowing, and texture