PhotoSurrealism : an object-based multi-projection camera model

圖像超現實主意 : 物件導向多重投影方法

Student thesis: Master's Thesis

View graph of relations


  • Chi Wai Stephen TSUI

Related Research Unit(s)


Awarding Institution
Award date16 Jul 2007


Photorealistic, as was the ultimate goal for most Renaissance artists, is not the main concern for most contemporary artists and that should come as no surprise to us. Beginning with Cubism, artists have been struggling to represent views of the world through their imagination. The development of art has been moving away from pure realistic representation and symbolism, towards artistic activities that involve the creation of something surrealistic and new. As one of the quantum physicists and philosophers, Bohm [1] points out: “It seems very interesting that the development away from representation and symbolism and toward what may be called ‘pure structure’ that took place in mathematics and science, was paralleled by a related development in art. Beginning with Monet and Cezanne and going to the Cubist and to Mondrain, there is a clearly detectable growth of realization that art need not represent or symbolize anything else at all, but rather that it may involve the creation of something new –‘ a harmony parallel to that of nature’ – as Cezanne put it. ” Similarly, being in the post-photographic era today saturated with highly vivid computer imageries, many artists have been dedicated to explore new forms of visual experience. The classical perspective projection, which provides much similarity to what we consider as normal vision, has long been served as the paradigmatic camera model in computer graphics for greater realism. By conforming to the standard shading model but altering the optical model of an ordinary Ray-Tracing algorithm, we propose an experimental method of generating computer images. The proposed model is titled ‘Object-Based Multi-Projection Camera Model’. The main reason is that individual object in a scene could be rendered by a unique camera model respectively. By mixing and matching various types of camera models in our multi-projection system, the potentiality of achieving dramatic visual outcome is immeasurable. Moreover, as the rendering process is initiated by constructing a rendering algorithm, for which the result is examined subsequently, the visual outcome of a final image is usually not foreseeable. Artworks created in this way could be highly experimental, and the outcome would often surprise not only the audience but also the creator. Through an archive of images and animations created by our proposed model, we discuss the characteristics and deviations of several non-standard camera models in comparison with the classical perspective. Moreover, by explicating the implementation of our experimental model, the rationality and accuracy of our algorithm will be illustrated. Last but not least, we further discuss the extensibility of multi-projection in comparison with single projection system, as well as the unpredictable outcomes create by our proposed method.

    Research areas

  • Digital techniques, Image processing, Computer graphics