5/19/2006

Professor Rahul Singh, Professor Ilmi Yoon

While 3D models are essential for computer graphics and scientific visualization, most modeling tools today are not easy to learn without special training and artistic skills. Even for a trained person, modeling process often involves labor intensive geometry modification steps. To make a modeling process more intuitive, effective, and fun, especially for domain experts who have knowledge of the subject to be modeled, I propose a new modeling method based on geometric parts, digital images, and sketch drawings. With this method, user can create various 3D models from a generic model by few simple steps: selection of body parts from a generic model, selection of target images, and sketch drawing on top of the image. Given these user inputs, proposed geometry morphing algorithm automatically deforms the generic 3D body parts into the user drawn 2D sketch while preserving the important characteristics of the original geometry. Therefore, the system allows user to construct desired 3D shapes only by a set of 2D inputs guided by the images. To improve photo-realism of the final 3D model, textures are also automatically extracted from images.

modeling, image, parts, sketch, geometry, morph, morphing, texture, OpenGL, Cg, shader, GPU