mental ray supports a first-hit rendering algorithm called rasterizer. It allows to achieve superior motion blur speed compared to the traditional scanline mode. The primary difference is the separation of the shading sample phase from final sample compositing (also called sample collection). This allows to tune for anti-aliasing and motion blur image quality independently of the shading computation. Without the rasterizer, mental ray is always shading each of the spatial and temporal sample points (eye rays) selected within the shutter interval. As a consequence, an increase of the anti-aliasing quality requires proportional raise of the shading effort.

The rasterizer works by shading objects at certain spatial and temporal sample positions, and caching these shaded samples. If an object moves, the shaded sample results can be re-used at each point of the motion blur path. The cache is tied to the geometry:

The late compositing of shading samples to form screen samples, and re-using of shading results has several important consequences:

The rasterizer can enabled with a scene option or on the command line of a standalone mental ray. mental ray controls the pixel over-sampling with the samples collect, which gives the number of samples per pixel-dimension. For example, the default value of 3 gives 9 samples per pixel. The rate of shading is controlled independently with shading samples, and defaults to 1.0, or roughly 1 shading call per pixel. Note, that this drives the internal tessellation depth, and takes effect after the geometry's own approximation has been calculated. It is possible to override the shading samples either per object or per instance.

For further acceleration of motion blur rendering, the shading frequency can be reduced for fast-moving objects by tuning the motion factor in the scene options or on the command line of a standalone mental ray. This is a positive floating point value which divides the number of shading samples taken, proportional to the speed of the motion of the object in screen space. The default value is 0.0, which disables this feature. A good starting value for many scenes is 1.0. Higher values will further reduce the number of shading samples thus raise performance for fast moving objects.

Due to the different sampling patterns it should be avoided to use sample passes rendered with the rasterizer together with passes that do not use the rasterizer.

Copyright © 1986, 2013 NVIDIA Corporation