Dead-line April 22nd 2005

Postdoc position in Non Photorealistic Rendering (NPR)

MIRO: Methods for the legible and Interactive Rendering Of 3D complex data

• Extension of Level-of-Detail techniques for legible rendering: When it comes to the rendering of complex scenes with large numbers of features, primitives, and rendering styles, choices have to be made in order to produce a legible solution. The issue is here similar to the classical issues of Level-of-Detail (LOD) management, but with a focus on legibility. We will experiment this notion of Legible Level of Details for the different visualizations and criteria developed during this project.

• Expressive rendering for legibility: Two main domains will be considered: (i) 3D rendering of objets characterized with feature lines and (ii) volume visualization using NPR techniques. For the feature lines, we will develop a geometric and photometric characterization of lines. For volume rendering, based for instance on data extracted with an X-ray scanner, complexity comes from the superposition of information. We will propose some algorithms for the extraction of interesting features inside those volumes.

• Simplification techniques for NPR: Visualizing a very complex scene (city, forest) or an object with many features, such as a tree, in a line drawing or hatching style produces too many lines. The result becomes soon incomprehensible due to line clutter. We will develop simplification techniques of the sets of lines to produce legible and interactive rendering. The reduction of density for strokes-based rendering will also be considered.