Point-Based Graphics

In recent years, point primitives have received growing attention in computer graphics. There are two main reasons for this new interest in points: On one hand, we have witnessed a dramatic increase in the polygonal complexity of computer graphics models. The overhead of managing, processing and manipulating very large polygonal meshes has led many researchers to question the future utility of polygons as the fundamental graphics primitive. On the other hand, modern 3D digital photography and 3D scanning systems facilitate the ready acquisition of complex, real-world objects. These techniques generate huge volumes of point samples and have created the need for advanced point processing. Conceptually, points constitute the atomic digital building blocks of object geometry and appearance - just as pixels form the digital elements of 2D images.

Our research in point-based graphics was inspired by the desire to design an alternative pipeline for efficient 3D content creation. Such a pipeline is displayed above. We start with a real world model which will be digitized by some 3D acquisition process. As a result we obtain at raw 3D point cloud. The first important component of our pipeline is a surface model - a powerful mathematical representation that interpolates the discrete point samples and allows for sophisticated surface processing including cleaning, hole-filling and the like. Next, users will want to edit the surface geometry or modify its appearance attributes. This includes editing operators, such as surface deformation and Boolean operations, as well as an editing metaphor. In addition, 3D painting, texturing carving and effect filtering is needed, in the same way as we know it from conventional 2D image editing. Other important tools encompass compression or digital watermarking. Finally, we have to provide efficient methods for the rendering of point sampled geometry to display our models in high quality. This includes dedicated point pipelines, hardware acceleration and the removal of aliasing.

During the past couple of years we have investigated the individual stages of this pipeline. We developed novel concepts for the representation of point sampled shapes including surface analysis, resampling, Fourier transforms and scale spaces. We also developed algorithms for interactive modeling of point clouds and for geometric processing of point models. We studied the potential of high performance rendering of point clouds, including advanced shading, antialiasing and transparency. To this end, we designed a powerful framework, Pointshop3D, which encapsulates the major components of a point-based 3D content creation pipeline and enables novices to easily start into the world of point-based graphics.

Currently, we devote research to the efficient compression of point models, to time varying point clouds and 3D video, to physics based modeling using points and to the design of point rendering and processing hardware.

	Feature Preserving Point Set Surfaces
	Dynamic Sampling and Rendering of Algebraic Point Set Surfaces
	Hardware Architectures for Point-Based Graphics
	Acquisition and Cleaning of 3D Objects
	Representation, Processing and Modeling
	Pointshop3D: An Interactive System for Point-Based Surface Editing
	Physics-Based Animation
	3D Video