毛髮繪製呈現之探討與研究

Abstract

在電腦圖學的領域中，毛髮的繪製呈現可以透過許多不同的方式去實現，其中包含了將模型透過材質轉換成類似毛髮的樣子、以及透過參考的引導線產生出一搓一搓的毛髮。每種方法各有優缺點，也能分別做出不一樣的效果與成果 人們總是不斷在追求更加真實的畫面呈現，由於目前硬體持續升級強化，電腦逐漸可以將複雜的場景與模型透過運算將畫面繪製得趨近照片般真實，同時又可以達到實時的等級。然而毛髮的數量十分浩大，光是一個人的頭頂可能就有超過十萬根的髮絲，假如想要將一根根的毛髮直接透過模型製作並繪製出來，那將要花上長久的時間與龐大的資源，也因此透過計算產生出毛髮的呈現方式繪製出如照片真實般的畫面便成了一個重要的解決方式 本論文透過研究數種毛髮與毛皮的實作方式，接著分析幾何著色器(geometry shader)的功能，以殼的生成法為基礎，透過幾何著色器能進行加點與變形運算的特性將模型的三角片複製並位移來產生出數層的殼，接著貼上使用隨機數生成的噪點貼圖來實現毛皮的生成，並以實時的速度呈現在使用者面前。Hair rendering can be implemented by many ways in computer graphics. For example, it can be implemented by transforming the model into hair with the textures. And it can also be implemented by using guide hair to generate hair strands. Each method can present different result and has its pros and cons. People always want computer graphics to be more realistic. Because of the improvement of the hardware, computers can gradually draw scenes and models that are more complex in real-time. However, the amount of hairs is too huge. There are up to hundred-thousands of hairs on a human head. Rendering hair with a model which has every strands of hair will take lots of time and enormous amount of resources. Therefore, we need to implement hair rendering by generating them with an alternative way. This thesis studies various methods to implement hair and fur rendering, analyses the features of geometry shader, and then implements the fur generation and modeling based on the Shells method. It uses geometry shader to replicate and transform the triangles of the models, and then to apply the texture which is generated by random numbers to implement fur in real-time.