The Geometry Processing Pipeline

The concept of a Geometry Processing or Reverse Engineering Pipeline can be found in most modern production industries. For example, in the automobile industry the rough shape of a car is typically first modeled in (real-life) clay and then transferred to a digital representation for further CAD processing. However, transferring the geometry to a digital form (e.g. by laser scanning) often introduces noise and inconsistencies, necessitating an optimization before further processing is possible. Also, the raw scan-data is typically unnecessarily complex and for a designer to be able to easily manipulate the model further, it needs to be brought to a more compact/intuitive representation (e.g. Spline/Subdivision). Finally, to be easily/cost-effectively manufactured, the final model needs to adhere to production technical or material induced constraints.

In this practical course we will visit the different stages of this pipeline. We will acquire real-life data using Microsofts Kinect, further process this data with state-of-the-art algorithms, and finally produce real-life models using cutting-plotters and 3D printers.

You will be divided into teams responsible for different, exciting parts of the geometry processing pipeline. Each team (2-3 people) will then implement the respective geometric methods and algorithms as plugins for the open source geometry processing and rendering framework OpenFlipper. In meetings on a regular basis we will provide the necessary support to ensure a good progress in all groups.

Requirements:

• C++ programming experience
• Depending on the project some OpenGL knowledge is preferable
• Preferable: Lecture "Basic Techniques in Computer Graphics"

Downloads

• Framework KinectView
• Slides Kick-Off Meeting
• Slides OpenFlipper Intro
• Slides OpenMesh Intro
• Example Models