Integrating Occlusion Culling and Hardware Instancing for Efficient Real-time Rendering of Building Information Models

This paper presents an efficient approach for integrating occlusion culling and hardware instancing. The work is primarily targeted at Building Information Models (BIM), which typically share characteristics addressed by these two acceleration techniques separately – high level of occlusion and frequent reuse of building components. Together, these two acceleration techniques complement each other and allows large and complex BIMs to be rendered in real-time. Specifically, the proposed method takes advantage of temporal coherence and uses a lightweight data transfer strategy to provide an efficient hardware instancing implementation. Compared to only using occlusion culling, additional speedups of 1.25x-1.7x is achieved for rendering large BIMs received from real-world projects. These speedups are measured in viewpoints that represents the worst case scenarios in terms of rendering performance when only occlusion culling is