Automatic Portal Generation And Path Planning In Virtual Reality Engine

The digital entertainment industry is developing very fast in China, but the key technology game engine is nearly occupied by foreign companies. So the research of game engine is important. Game engine is a special kind of virtual reality engine, concerning graphics, artificial intelligence and other research fields, and its complexity is growing rapidly.In this article, we first review several key aspects of game engine, and focus intensive introduction on two research fields: large-scale rendering and path planning algorithms. Then rendering acceleration techniques for large-scale scenes are introduced, including the main algorithms in visibility determination and level of detail. There portal technique for visibility determination is highlighted.Portal technique is widely used in virtual reality engine, but automatic portal generation remains a problem. In this article, we first explain the key point and difficulty of portal generation, and analyze the strong and weak points of previous works, then a brand new concept for cells validation is introduced, the InvisRatio field. We introduce an algorithm for generation and validation of portals, based on characteristics of distance field and InvisRatio field near portals. Our experiments proved that the algorithm can find the portals ignored by other approaches, and by checking InvisRatio, we can make sure of the validity of these generated portals.Path planning is also one key aspect in virtual reality engine. One of its application automatic navigation of virtual camera is a useful technique, and worth studying. But fully automatic camera navigation algorithm is seldom seen, or not so perfect. In this article we first review various path planning and camera planning approaches, and introduce some basic rules for camera planning. Then we introduced a systematic camera planning algorithm with four phases including path planning, path smoothing, trajectory planning and orientation planning. We use octree to decompose our scene and improve the efficiency through decomposing model, decreasing nodes count in connectivity graph, avoiding collision test etc. Ideas from visibility graph are used to help us optimize our path. The algorithm will quickly produce regular path segments with few turns, which fit for camera planning. And efficient path smoothing, trajectory planning and orientation planning algorithm are also introduced. Our experiments proved the final result obeys these rules. The path planning path of the algorithm can also be used for other 3D fly navigation application.At the end of the article, we summarize the strong and weak points of these two algorithms, introduced the possible improvement for them and evolving direction of these techniques.