油畫中厚塗技法筆觸風格之模擬

Abstract

在本論文中提出自動化系統能夠藉由非擬真顯像技術及加入虛擬光源將一張圖片模擬出具有厚塗風格的油畫。為了模擬實際手繪油畫效果，筆觸方面本系統參考梵谷畫家漩渦狀的筆觸特色並藉由邊緣偵測及Edge Tangent Flow演算法來決定筆觸方向。配合虛擬光源下所建立兩種不同筆刷,其效果在視覺上如顏料厚度深淺的變化。最後結合色彩簡化產生厚塗風格油畫。

In this thesis, an automated system is proposed for transforming an input color image into an oil-painting in which impasto techniques are heavily exploited in whirling-like brushing. The transformation goes through three major functional stages as follows: 1. dynamic color quantization in RGB space. 2. context establishment for generating brush strokes with whirling-like visual effects. 3. context establishment for generating visual effects resulting from the use of impasto painting skill.

Judging by the results obtained from a series of image tests, the system is capable of generating desired visual sensation in the simulation as was initially expected.