General Purpose Visibility

Visibility culling is the task of determining the visible geometry from a view point or a region. It is considered as one of the classical problems in the field of computer graphics and visualization. Regardless of the rapid advances of hardware performance and performance of consumer graphics hardware in particular, visibility culling is of immense importance for real-time rendering applications like computer games or architectural walkthroughs, as the complexity of the models and the amount of data (e.g., consider applications like Google Earth) grows accordingly. Apart from the traditional use of visibility culling for rendering acceleration, visibility culling is required for many other important applications, e.g., for surveillance, robotics, or line-of-sight computations in complex simulations. Amazingly, there are virtually no visibility solvers available that are robust, fast, and work for general scenes despite the long history of visibility research. For example, a large amount of time and effort during game development is spent for visibility computations, because visibility culling has to be done fully or partially by hand, a task that ought to be fully automatic. We believe that the reason for this situation lies in the fact that there are still fundamental research issues left unsolved. In this project we want to tackle these issues, and provide solutions that work for general scenes without restrictions to size or type of the scene, whether it is a city scene or a forest scene containing complex vegetation (a notoriously difficult case for any visibility culling method). We will investigate a variety of visibility culling problems, because we think that all of them have their own share of important applications. We will propose methods for visibility preprocessing in a novel progressive and global manner, methods for online occlusion culling that also work for dynamic scenes, hybrid methods that combine visibility culling with rendering simplification for massive scenes with hundreds of millions triangles, as well as methods for interactive visualization of the visibility properties of a scene. We believe that this project will provide valuable contributions in many areas of visibility re-search, and that our powerful concepts have high potential to become very useful in practice. This project may as well bring the final closure to the topic, necessary to make visibility feasible for general purpose.

Visibility culling is the task of determining the visible geometry from a view point or a region. It is considered as one of the classical problems in the field of computer graphics and visualization. Regardless of the rapid advances of hardware performance and performance of consumer graphics hardware in particular, visibility culling is of immense importance for real-time rendering applications like computer games or architectural walkthroughs, as the complexity of the models and the amount of data (e.g., consider applications like Google Earth) grows accordingly. Apart from the traditional use of visibility culling for rendering acceleration, visibility culling is required for many other important applications, e.g., for surveillance, robotics, or line-of-sight computations in complex simulations. Amazingly, there are virtually no visibility solvers available that are robust, fast, and work for general scenes despite the long history of visibility research. For example, a large amount of time and effort during game development is spent for visibility computations, because visibility culling has to be done fully or partially by hand, a task that ought to be fully automatic. We believe that the reason for this situation lies in the fact that there are still fundamental research issues left unsolved. In this project we want to tackle these issues, and provide solutions that work for general scenes without restrictions to size or type of the scene, whether it is a city scene or a forest scene containing complex vegetation (a notoriously difficult case for any visibility culling method). We will investigate a variety of visibility culling problems, because we think that all of them have their own share of important applications. We will propose methods for visibility preprocessing in a novel progressive and global manner, methods for online occlusion culling that also work for dynamic scenes, hybrid methods that combine visibility culling with rendering simplification for massive scenes with hundreds of millions triangles, as well as methods for interactive visualization of the visibility properties of a scene. We believe that this project will provide valuable contributions in many areas of visibility re-search, and that our powerful concepts have high potential to become very useful in practice. This project may as well bring the final closure to the topic, necessary to make visibility feasible for general purpose.