Skip to Main Content
We introduce an efficient rendering technique using temporal coherence to make a 3D scene including complex but static 3D models. In order to render a realistic and complex 3D scene, we need to use huge amounts of polygon or complicate rendering algorithm. Almost computer graphics techniques have been researched to generate sophisticated image in real-time. Since consecutive image sequence has temporal coherence, a previous render result is a good resource to predict a current scene. We replace static objects with previous frames which are drawn for static objects. Fixed viewport allows us to reuse previous frames without any modification. However, sometime a static object becomes dynamic and we have to update previous frame. We reduced the update overhead by distributing static objects in several group. Our method generates static object buffers as many as the number of group and updates each buffer independently. As a result, we can present tens of millions of polygons in real-time.
Date of Conference: 26-27 April 2012