Culling using linear bounds for stochastic rasterization

摘要

We present a new culling test for rasterization of simultaneous depth of field and motion blur, which efficiently reduces the set of (x, y, u, v, t) samples that need to be coverage tested within a screen space tile. The test finds linear bounds in u, t space and v, t space respectively, using a separating line algorithm. This test is part of the foundation for an efficient 5D rasterizer that extracts coherence in both defocus and motion blur to minimize the number of visibility tests.

主权项

1. A method comprising:
using a processor to form separating lines between moving triangle vertices and a screen space tile; using said separating lines in ut-space and vt-space to cull samples for graphics processing; and performing, in a processor, stochastic rasterization of simultaneous motion blur and depth of field using samples that were not culled.