影像式漫畫肖像生成之研究

Abstract

被譽為第九藝術的漫畫，不論是在電影、電玩等娛樂產業中，佔有相當重要的地位。透過線條的勾勒以及漫畫家獨特的筆觸運用，呈現其在漫畫劇情中所想表達豐富的內涵與視覺效果。而其中對人物畫像細膩的風格描繪，更受到現代漫畫迷所喜愛。

在電腦圖學領域中要產生具有特定藝術風格的人像，具有相當高的挑戰性。而本論文則是希望透過電腦創造、模擬具有藝術價值的作品，其過程是經由輸入一張真實的人臉照片，運用本論文所提出的NPR技術轉換成具有特定漫畫家畫風的人物肖像。在實作上我們根據漫畫家井上雄彥的畫風做為參考，根據繪畫元素分成三個部分技術來做NPR處理，分別為臉部描繪技術、頭髮描繪技術，以及影線描繪技術。最後合併三部分結果做為輸出，得到具有特定漫畫家畫風的人物肖像。

The Comics, Ninth Art, gets more and more important in the entertainment industry, like movies and games. The comic artists use the particular brushwork and contour to express the rich content and visual effects. Especially the exquisite portrait painting that comics fan is spellbound.

Creating a portrait in the style of a particular comic artistic is a difficult problem. In this paper, we introduce an NPR system to generate Human Portrait in Comic style. From observation, we found out that there are three parts of element in Comic Art style:1) facial contour, 2) hair component, and 3) hatching of shadow. So our system process these part individually, make these three parts are stylized in Comic style. At first, we input a human face photo, and the feature point of human face. And then we simulate the comic artist painting style in the three parts. Finally, users may generate a Comic style portrait easily by using our system without any painting skill.