Splat Based Raytracing

Splat based raytracing refers to the use of raytracing techniques to visualise point models. Point models have surfaces represented by unconnected points, rather than triangles. The use of triangles for rendering is efficient as long as the model complexity is low, and each triangle contributes to a significantly large area of the screen. As triangle based models become increasingly complex, the individual triangles become smaller and smaller, till each triangle contributes to a single pixel on the screen. At this stage, the overhead of using triangles to represent the model is unjustified. Point based models are efficient alternatives for representing such complex objects because they do not maintain connectivity information between primitives. Raytracing is a rendering technique that simulates the behaviour of rays of light in a scene. It is widely used in polygonal scenes to simulate realistic lighting models. The use of raytracing greatly enhances the realism of rendered output. In this report, we discuss the concept of point models and splats, and how existing raytracing techniques can be extended to splat based models. This includes the construction of splats from point data and ray-splat intersections. We also examine how the Phong shading model (Phong splats) can be used to improve the normal vector estimate at the point of intersection.