三維簡化動畫模型之動態輪廓裁剪

Abstract

為了減少投影複雜三維動畫模型的時間，以求維持投影時間能夠及時，我們提出以簡化動畫模型套用材質貼圖以及輪廓裁減的方式來逼近原三維動畫模形的外觀。為了紀錄複雜三維動畫模型的法向量資訊，在前置運算中我們建立了法向量地圖，並且在執行運算中將法向量地圖轉為適合簡化動畫模型的材質貼圖。在執行運算中，我們也運用前置運算中分割完成的法向量空間集合來加速尋找複雜三維動畫模型的三維輪廓。運用構成這些輪廓的邊，我們利用OpenGL中的模板快取﹝Stencil Buffer﹞把已投影簡化三維模型中多餘的部份截除。最後我們能以較快的投影速度達成較高的影格更新率﹝Frame Per Second﹞。

Considering a sequence of detailed three-dimensional polygonal models, we develop a system framework to approximate them using a pre-computed sequence of coarse three-dimensional polygonal models alone with normal maps. We use the normal maps as information for reconstructing the surface appearance of the original detailed polygonal models. During the runtime of rendering the coarse model sequence, we introduce a simple clustering method to extract the silhouette of detailed three-dimensional polygonal models. We cull the coarse polygonal models by using the silhouette found before to approximate the 2D contours of detailed polygonal models. In order to cull the extra pixels rendered off, we utilize the stencil buffer in OpenGL. To set the stencil value, we also introduce the stencil setting algorithm.