Using Local Information for Compositing CG Into Traditional Art

Abstract

I describe a procedure for compositing digital graphics into traditional artwork based on local characteristics of the art. This is based on understanding that variations in pictorial characteristics such as perspective, lighting and color, or style naturally occur in many examples of traditional artwork. The goal of this study is to create composites that are visually believable while showing that an object composited into one section of an image requires characteristics designed speci cally for that section. In order to show this, I examine four di erent case studies. Each case is a work of traditional art which I composite di erent computer graphic elements into. These CG elements range from simple primitive objects to complex character models.