Radiosity on evolving networks (Univ. Dortmund) (Q2726308)

This small book treats of digital image synthesis using neural networks and finite elements: growing cell radiosity.NEWLINENEWLINENEWLINEIt presents an approach combining Monte Carlo sampling with finite elements scheme in order to solve the radiosity equation.NEWLINENEWLINENEWLINEThe main advantage of the method seems to be that it does not rely on prior knowledge of surface or geometric objects present in the scene. The triangulation of the scene can be derived as a byproduct of the final computations.NEWLINENEWLINENEWLINEThe main problem stated by the author is that the method ideally deals with smooth goal functions and is not well suited to handle high variances in the light flow.NEWLINENEWLINENEWLINEThe booklet is divided in 3 main chapters. After an Introduction surveying the Radiosity Illumination problem, chapter 2 describes the growing cell structure that is the core of the method. Chapter 3: Radiosity with cell structures, specializes chapter 2 for the radiosity problem and provides FEM integration model.NEWLINENEWLINENEWLINEChapter 4 is about applications but stays at a quite abstract level. The synthesized scenes are not very convincing.NEWLINENEWLINENEWLINESome conclusion and appendices are also provided.