可见性驱动的复杂场景连续多分辨率绘制

提出一种视点相关的连续多分辨率模型与可见性剔除相结合的复杂场景加速绘制算法。将复杂场景表示为场景图,然后自底向上构建场景的连续分层层次细节(CHLOD)模型;在实时绘制时,利用图形硬件的遮挡查询功能实现快速的层次遮挡剔除,并根据遮挡查询返回的可见象素数近似度量物体的可见性程度,通过与传统视点相关的细化准则相结合,实现了一种可见性驱动的场景连续多分辨率简化绘制方法,能够在保证场景绘制精度的同时进一步提高场景绘制的速度。

Visibility-driven Continuous Multiresolution Rendering of Complex Scenes

An accelerated rendering algorithm was proposed for 3D complex scenes that combined both the view-dependent multiresolution model and the visibility culling technology. In the preprocessing, the scene was represented using a scene graph and then continuous hierarchical levels of detail (CHLOD) for each scene graph node was computed. At run time, a novel hierarchical occlusion culling approach based on hardware occlusion queries was proposed and the number of pixels visible in the bounding box was used to estimate the visibility of each scene graph node. By combining the visibility with the traditional view-dependent criterion, a visibility-driven continuous multiresolution rendering system was implemented. The experiment results show that the algorithm can increase rendering speed while keep the image quality.