The reason is there's always some numerical fuzz in calculating the intersections, and if the point is not offset, the shadow ray might hit the very same polygon again, putting shadows in wrong places. When the intersection between the eye ray and the scene is found, the intersection point is moved (offset) towards each light before calculating the shadow ray. The spotlight grid is defined in number of grid cells instead of pixels, because there is no pixel size associated with lights. This speed increase only applies to spotlights, because they are very similar to the viewports, the light location is like the camera location, and the light cone defines the viewport. To increase calculation speed for the spotlight shadow rays, the render plug-in divides the spotlight cone into rectangular regions, and again builds a sorted list of the objects within each region. Usually the lions share of the render time is spent tracing the shadow rays. The shadow ray is the ray shot from the scene towards each light when the intersection between the eye ray and the scene is found. The smaller the grid cell size, the more memory it takes and the more memory it takes to build, but the faster the final render. The width and height of each grid cell in pixels. Advanced Rhino Render Settings Render acceleration grids Screen grid, cell size
