具觀視點相依多層次精細度模型保留之即時且連續的地形顯像技術

Abstract

論文摘要 即時且平順地顯像大規模的地形資料是虛擬實境中重要且深具挑戰的問題。為了達到有效的地形顯像，本論文嘗試用動態產生的多層次精細度模型，在不影響顯像品質下儘量化簡地形網格（或稱為地形模型）的三角形數目。在地形即時顯像系統內，我們設計了溝渠式的區間分割，有效地解決過去一般的分割方式在區間邊界上，因多層次精細度選取的不一致而產生幾何不連續的現象。我們也提出保留可延續使用的多層次精細度模型之方法，並群組相鄰近的地形頂點來計算高度差投影值，有效地利用時間與空間的共效性，在犧牲極少量，甚於可忽略的影像精確度下，獲得高度的顯像加速，使得地形顯像更為即時與逼真。此外，亦應用地形區間具有規則的特性，大量縮短可視範圍擷取的計算與多層次解析度貼圖的時間。總之，本論文整合多層次精細度模型的顯像架構，解決一些過去無法克服的問題，並建立地形顯像系統的核心，適合一些以地形顯像為基礎的應用，諸如飛行模擬、地理資訊查詢系統等。

Abstract Real-time and smooth rendering of a large-scale terrain data has been a challenging problem. In this paper, we propose a geometrically continuous view-dependent level-of-detail (LOD) modeling aiming to speed up the generation of terrain mesh and in the meantime achieve a satisfied image quality. The terrain data is subdivided into blocks and each of which will possesses its own LOD mesh that is dynamically determined according to the viewing parameters. Between two adjacent blocks, a dike structure is proposed that aims to provide a smooth blending between two meshes of different levels of detail, and hence remove cracks that usually occur in previous methods. We also propose a mechanism in LOD modeling that caches the LOD of a block for the possible reuse in the following frames after it is generated. Since LOD selection and generation in general requires computation on each node level, such a LOD caching can potentially contribute a considerable saving of computation time. We also implement several speed-up methods, such as view culling and texture mapping, and others that utilize spatial and temporal coherence. The proposed view-dependent LOD modeling techniques will be useful in building a terrain rendering system that has applications in the areas such as flight simulation and GIS.