基于OGRE的大规模场景核外渲染的研究

A Research of Out-of-Core Rendering of Large-Scale Scenes Based on OGRE

传统视景系统中采用所有模型全部加载进内存的方式,不适用于超过了内存大小的模型的渲染,也严重影响实时性。针对此问题,本文首先将大型场景模型进行预处理,切分成适合核外数据加载的较小模型,然后通过视点预测、模型预加载、视域外模型的卸载等操作实现超大场景模型的实时渲染。经实验验证,该方法可以实现超过内存限制的大场景模型的实时绘制,效率较高。

The real-time rendering of large-scale scene is a key issue in the visual system of virtual reality,digital city and train driving simulator.With the loading of all the models into the memory,the traditional visual system is not appropriate for the exceed-memory-capacity models＇ rendering,and seriously affects the real-time effect.This research first pretreats large scene models,cuts out-of-core data for loading into smaller models,and then achieves the real-time rendering of large-scale scenes through the prediction of view point,the model pre-load and the out-of-vision model unload.The experiment shows that this method makes the real-time rendering of large-scale scene models with insufficient memory possible and with higher efficiency.