Portal rendering

The portal -based rendering is essentially different from other algorithms since it was designed for use by interior views (indoor scenes). In contrast to spacious areas usually only a very small area to render out of the question. Background of the method is the following idea: The environment consists only of spaces that are connected by so-called portals (doors, windows, etc.). So come for the visibility region only the space in which the viewer is, as well as all adjacent spaces into consideration (recursively). The room layout is not part of the process, but is pre- set by either designers or calculated by automated algorithms.

First of all polygons in the room where the camera is tested for visibility and rendered. Then it is checked whether there is a portal in the visibility region, and the space is located behind the portal, rendered with a correspondingly reduced visibility area. This process is recursive, that is, he calls itself until all visible rooms have been processed.

An advantage of this rendering is that you can easily link it actually geometrically independent spaces together. Thus, for example, a mirror generated by setting specifying twice the same space for a portal. Even the otherwise very complex calculation of light and shadow effects is easier here: First, a scene with the visibility range of the light source is calculated (not really rendered! ) And lighted accordingly. Subsequently, the scene is rendered from the perspective of the viewer.

Since many steps such as the visibility test can be performed in hardware, the implementation of such an algorithm is easy and the performance is due to the hardware acceleration quite respectable. The major disadvantage of the portal -based rendering, however, is that it is entirely inappropriate for outdoor scenes. Therefore, this algorithm is combined in many modern 3D applications with other methods such as Binary Space Partitioning or octree.