摘要
一般大眾在觀賞電影時，所得到的情緒反應有絕大多數的誘因是來自於電影中角色的表演，相較於觀眾透過場景、音效等等所得到情緒訊息則相對較少，而電影中的人物不管是真實人物或虛擬角色，在傳達給觀眾的情緒訊息當中，表情的表演仍然是提供最多訊息管道。由於動畫在數十年來的進步與龐大的產量，虛擬人物在情緒的要求也就越多而且更細膩，動畫師在製作表情動畫的時候則必須先收集相當多有關表情的資料來做參考，這樣的製作過程需要相當多的時間成本。
本研究是利用形變(morphing)的技術，並假設角色的表情亦有最單純”元素”的想法，假設人類基礎表情的九個模型，其中九個基礎模型分別為：哀傷、悲痛、生氣、驚訝、嫌惡、輕蔑、快樂、笑、害怕，在定義九個基礎表情後，我們利用3D繪圖軟體Maya來建構三維模型，並利用Virtools軟體來製作可供操作的模擬介面，介面中包含兩個可控制的區域：(1)顯示區的控制(2)操作區的控制，模擬介面中使用者可調整模型在顯示區的角度，並可利用操作區來控制基礎模型在內插值中的變化程度，在模擬介面中的取樣最少可做單一基礎表情的形變程度，最多則可操作九個基礎表情的參數，並在顯示區中的模型可立即顯示形變後的結果，我們希望這樣的方法與構想能夠更快速且準確的提供動畫製作時更多的參考資料。

Abstract
When the general public watches movies, most of inducements of emotional reactions obtained are from actor performances. Backdrops and sound effects, on the other hand, carry far less emotional messages. However, whether characters in movies are actual persons or virtual roles, performance of facial expression is still the primary method to provide the most messages to audiences. Due to the fact that animation has been improved and was in mass production for decades, the requirements for virtual roles’ emotion expressions are more fine and delicate. When animation makers create facial animation, they need to gather a considerable amount of data related to facial expressions as reference, and this production process requires the consumption of large quantity of time cost.

This research uses the morphing technique and theorizes that characters’ facial expressions contain the purest “elements,” which assumes that there are nine models of human fundamental facial expressions: sadness, grieve, anger, surprise, hatred, scorn, happy, laughing, and fear. After defining these nine fundamental facial expressions, we can use 3D Graphics Software Maya to compose 3D models, and use Virtools Software to create an operable simulation interface which comprises of two controllable areas: (1) control of display area, and (2) control of operation area. A user can use the simulation interface to adjust angles of models in the display area and to control variation of fundamental model models in the interpolated value by using the operating area. At least one single fundamental facial expression in the morphing degree can be accomplished through samplings of simulation interface, and up to nine parameters of fundamental facial expression can be operated. Also, models in the display area can immediately show results after morphing. We hope these methods and ideas can rapidly and precisely provide more reference data during animation process.

一、中文
1.保羅．艾克曼著，易之新譯2004，心理學家的面相術，心靈工坊，台北。
2.崔小萍，1984，表演藝術與方法，書林出版有限公司，台北。
3.詹竹莘，1997，表演技術與表演教程，書林出版有限公司，台北。
二、英文
1.Arad, N. & Reisfeld, D. (1995) Image warping using few anchor points and radial functions. Computer Graphics Forum, vol.14, pp.35-46.
2.Apostolos, L., Chase, D. G. and Marc, L.(1995). Feature-Based Volume Metamorphosis, In Computer Graphics Proceedings, Annual Conference Series, ACM SIGGRAPH, pp. 449-456.
3.Ed Hooks.(2003). Acting for animators: a complete guide to performance animation. HEINEMANN Portsmouth, NH.
4.George, W. (1990).Digital Image Warping. New York: IEEE Computer Society Press.
5.Kanai, T., Suzuki, H., Kimura, F.(1998). Three-dimensional geometric metamorphosis based on harmonic maps. The Visual Computer 14,4,166-176.
6.Marc, A., Daniel, C., David, L.(2000), As-Rigid-As-Possible Shape Interpolation, ACM SIGGRAPH, pp157-164.
7.Richard, G. B., Rahul, J & Justin, R.(1999). Digital Image Processing: Image Morphing. Rice University. November January 12, 2006, from the World Wide Web: http://www-dsp.rice.edu/courses/elec539