如果說繪畫「是」一個世界(a painting is a world)、攝影是「關於」一個世界(a photograph is of a world)、電影則是「為了」一個世界(cinema is for a world)；那麼，電腦只是在一個世界「之上」(the computer is simply on a world)。如今，影像的生成主要是透過二元的編碼與演算法來完成，就科技面向而言，視覺不再依附於任何影像的物質性(materiality)上。如今，在後數位世代的影像創製實踐中有兩個極其突出的標準：一是在對話脈絡(dialogical contexts)裡，不同文化與媒介元素之間的跨接(crossover)；二是在開放結果過程(open-ended process)指涉互動性以及虛擬媒體應用軟體中不同觀點、態度與真實之間的互動。在這些領域裡，人們可以經驗到多變性與多元性，甚至超越主流的商業、產業的影像與媒介生產模式。為了討論超越「影像」概念的藝術實踐，本研究試圖從當代對於「後」影像(Post-image)的觀點出發，並援引當代影像藝術創作作為實踐案例探討諸多科技—美學—藝術策略，以及較廣義的創作實踐，談論影像創製的美學問題。最後，本文亦將針對當前結合深度學習與偽造而成的Deepfake人工智慧人體圖像合成技術，談論後數位時代影像從拼貼到偽造所造成的政治性與影響。

If a painting is a world, a photograph is of a world, and cinema is for a world, in that case a computer is simply “on” a world. Nowadays, images are generated mainly from binary code and algorithms. Recent technological advances have made the materiality of images no longer a prerequisite for vision. There are two extremely prominent features in the post-digital generation‟s practices of image production today. The first is the cultural- and media-crossover in dialogical contexts; and the second is the referential interactivity in open-ended processes and the interplay among different views, attitudes and realities via virtual media software. In these fields, we can experience variability and diversity, and even break free from the shackles of the mainstream mode of producing commercial, industrial images and media. Treating the contemporary perspective of “post-image” as the point of departure and undertaking a case study of contemporary image art creation, this research is a three-pronged investigation into multiple technology-aesthetics-art strategies, creative practices in a broader sense, and aesthetic issues concerning image creation, insofar as to discuss artistic practices beyond the concept of “image.” This research is also intended to assess the political implications and consequences of image collage and forgery in the post-digital era with a special focus on Deepfake, an AI-based human image synthesis technology that integrates deep learning with faking.