In order to further improve the efficiency and reality of large-scale terrain rendering,an adaptive subdivision algorithm using GPU is proposed. The traditional method which tessellation occurred in vertex shader,needs a large number of tessellation templates and complicated treatment to cracks. Subdivision process of terrain in real time interactive process is not very efficient. To get the coarse grid of terrain,persistent grid mapping method is used. In the tessellation control shader,with screen projection error metric of interpolation point,each side of triangle is seamless adaptive subdivided. In the process,use the smooth interpolation value of subdivision level measure to modify calculation process of error,ensuring the monotonicity of error measurement. Finally,texture map is mapped to terrain using texture coordinate in the fragment shader. Experimental results show that the algorithm improves the density distribution of the grid,which is more effective than traditional adaptive subdivision method.%为了进一步提高大规模地形渲染的效率和真实感,提出一种利用GPU细分着色器进行自适应细分的LOD地形算法. 传统细分方法在顶点着色器中进行,需要预先计算细分模板且裂缝处理较为复杂,在实时交互过程中地形的细分效率并不高. 本算法首先利用固定网格投射的方法得到地形的粗糙采样网格,节省了视锥体裁剪过程,并且减少了裂缝出现的可能性. 其次,在细分控制着色器中利用插值点的屏幕投影误差作为误差度量方式,不断逼近误差阈值. 在此过程中,采用细分等级测度的平滑插值对误差计算过程进行修正,保证了误差度量的单调性. 最后,基于地形三角形各边的细分等级进行网格三角形无模板的无缝自适应细分. 实验结果表明,算法改善了网格的密度分布,与传统细分方法相比效率更高.
展开▼
