程序式表面瑕疵合成之研究：金屬鏽蝕的模擬與描繪

Abstract

時至今日，擬真描繪技術（Photorealistic Rendering Technique）已經發展到能夠產生令人印象深刻的電腦影像，但是擁有逼真的光影效果還是不夠，因為電腦影像中完美無暇的物體表面總是讓人可以很輕易地分辨出何者是真實照片何者是電腦影像。近幾年來，一些電腦圖學的研究學者也逐漸了解到表面瑕疵的出現是讓人感受到畫面真實的一個重要因素。在本篇論文中，我們使用一種程序式模型（Procedural Model）來模擬埋藏於地底的千年古代青銅器以及沈沒於深海中的鐵製品它們外觀上與成份上的演變。 埋藏於地底幾千年之後，古代的青銅器物會因為老化現象（Aging Phenomenon）以及不同土壤的物理和化學性質而在表面上產生各式各樣的銅綠，同樣的，在深海中的鐵製品也會展現出不同型態的鐵鏽。在本篇論文中，我們採用一趨勢值（Tendency）來整合所有會影響金屬□蝕的因素。在太陽與月亮的重力牽引下，潮汐與洋流使得海底的環境比地底要來得複雜與多變，在模擬海中的鐵鏽時，我們額外加入一個簡單的水流模型來模擬海水週期性的流動變化對鐵鏽的影響。此外，更進一步考慮到鐵鏽生成對周遭環境的改變。 我們所提出的模型包含兩個主要的模組：一個是環境模組（Environment Module）另一個是繁衍模組（Propagation Module）。環境模組以不同的趨勢分佈來描述特定環境下不同的金屬□蝕發生的可能性，它整合了器物本身的幾何性質以及環境的外在因素；繁衍模組本身是一個分支系統（L-Systems），根據環境模組提供的訊息來模擬不同的□蝕產物在金屬表面的生成與擴散。 利用改變環境模組的趨勢分佈，我們的模型可以產生各種不同的金屬□蝕結果，而且能夠擴充此一模型去模擬不同種類的金屬其□蝕的過程與外觀上的變化。我們也同時應用此模型來模擬中國四川三星堆文化的古代青銅器上的銅綠。

Today's photorealistic rendering techniques can produce very impressive images. But the rendered objects are often too clean and too shining. However these rendered objects are looked unnatural. Computer graphics researchers have recently revealed that the presence of imperfections is an essential part of the illusion of photorealism. In this dissertation, we propose a physically-based model to simulate the appearance and the development of patinas in the soil environment and rust in the sea. Been buried in soil over thousands of years, the ancient bronzes are covered by various patinas which are resulted by the interaction between the soil environment and bronze surfaces. A tendency is used to represent the accumulative effect of all factors on corrosion. The sea is more complicated than the soil. The tide and the current, under the gravity of the sun and the moon, affect the stability of the environment at the bottom of the ocean. A simple current model is employed to elucidate the periodic transition of currents in the ocean. Feedback to the environment of newly formed corrosion products is also considered. The proposed model consists of two modules – an environment module and a propagation module. The environment module adapts tendency distributions to reflect the potential of corrosion in a specific environment. The propagation module, itself an L-system, modulates the diffusion of corrosive products according to tendencies based on the object geometry and environmental factors. Changing the underlying distributions of tendency allows images of a variety of metallic corrosion to be generated. Besides this model can be extended to simulate a variety of metallic patinas and rust. We also apply our proposed approach to the ancient Chinese bronzes founded at San-hsing-tui.