Accelerated Graphics Hardware Rendering

mental ray 3.3 and later support hardware rendering. Hardware shading refers to leveraging hardware graphics acceleration boards installed in PCs or workstations, available from vendors such as NVIDIA, ATI, 3Dlabs, Matrox, SGI, Sun, and others, that can draw 3D geometry at very high speed. This document uses the term "hardware" to refer to 3D graphics hardware.

3D graphics hardware is not accessed by programs directly. An intermediate 3D library is used. Two standards exist: OpenGL and DirectX, both of which provide roughly equivalent functionality. However, DirectX is not portable and available only for Microsoft Windows, while OpenGL is available for a large range of platforms other than Windows that are supported by mental ray. Since one of the primary applications of mental ray is its use in large networks ("render farms") and such networks almost never use Windows for various technical reasons, OpenGL was chosen for mental ray to access graphics hardware.

mental ray versions prior to 3.3 already supported OpenGL rendering for a limited form of hardware acceleration, primarily for 3D projection and visibility calculations for rendering and shadow mapping. The scene was painted using OpenGL, but not with its true surface colors but pseudo colors that encode addressing information that tells mental ray for every pixel which object and which triangle is in front. mental ray then only needs to perform shading, but not intersection calculations. For actual color rendering, the visual quality of hardware shading was not acceptable so this was done in software.

Hardware capabilities have grown significantly since then. It is now possible to perform hardware shading with a simple form of procedural shaders. Shaders allow visual effects that go far beyond the "plastic look" limitations of older designs. It is now becoming practical to use graphics hardware for actual color shading at very high speeds without compromising image quality in unacceptable ways. mental ray 3.3 is designed to implement this new approach.

home

<< prev

next >>

contents