This project endeavors to answer the fundamental research question whether it is possible to alter the appearance of specific reflectance properties in a photograph of a scene by only partially characterizing the appearance of the scene via measurements. Specifically, the research team investigates novel appearance editing techniques that minimize the modeling effort, while maintaining maximum editing flexibility and at the same time impose minimal restrictions on the scene, while guaranteeing a physically plausible editing result. A common application of appearance modeling is to digitally clone a real-world scene, then change some scene properties, and finally revisualize the scene. Despite the enormous advances in appearance modeling, it still requires significant effort and expertise to create a digital clone of a physical scene. The effort needed to make the desired alterations to the scene often pales in comparison to the required appearance modeling effort. This project focuses on three reflectance phenomena suited for measurement-based editing: translucency, surface albedo, and surface reflectance.
This research program has a far-reaching impact beyond computer graphics. It has the potential to impact product development and pre-visualization, virtual reconstruction of archaeological artifacts, forensics, virtual cosmetics, fine-arts and entertainment, and other applications that require the visualization of an object/subject with altered reflectance conditions. Besides supporting the traditional research roles of graduate students, undergraduate education is enhanced by hands-on research experience and the incorporation of research results into existing courses.
