mental ray supports to store positional data with custom properties in the scene database for various purposes. This allows to implement rendering effects which are not attached to a surface but to a point in space (or time), like global illumination data. Custom applications and shaders may use it to compute and render particle systems or fluid density grids with mental ray. Most of the internal point data structures in mental ray which are used for caching purposes like the photon map are stored in proprietary formats, giving the freedom to adjust their behavior at any time in order to add new features or improve performance. The map data container, on the other hand, is a general data storage for point-based data, and designed to be suited for exchange of public information within the same or with other similar applications.
The map data container allows to store point-like information in mental ray. It provides support for automatic caching of data and swapping to disk files, to be able to manage arbitrarily large amounts of data.
The position of a point, the primary and mandatory property, can be specified in 1-dimensional up to 6-dimensional space, typical applications may use the 2D or 3D space. Any number of additional data item can be attached to a point or particle, each identified by type and name. This makes the map data container a perfect candidate for passing information between renderings, like between frames of an animation, or even from and to other rendering solutions. In addition, mental ray is optimizing common access and lookup methods with the help of acceleration techniques that are optimized for sparse and non-uniform distributions of the data, like arbitrary clouds of points.
mental ray supports to render point clouds as particles with a predefined shape, like sphere or tube. The particle information is stored as map data with certain additional properties, for example to control the size of the shape per particle. The rendering engine will automatically benefit from the underlying caching and acceleration system for the map data, thus being able to handle millions of particles efficiently.
Copyright © 1986, 2013 NVIDIA Corporation