Rasterization

Rasterization projects each triangle in the scene onto the screen and works out which pixels it covers. Those pixels are then shaded using material and lighting information, while a depth buffer (z-buffer) makes sure the nearest surface wins. Graphics cards are built for exactly this job, so the process is extremely fast.

Strengths and weaknesses

Rasterization treats each triangle on its own and has no direct knowledge of how one surface lights another. Shadows, reflections and bounced light therefore need extra techniques: shadow maps, screen-space reflections, ambient occlusion, pre-baked lightmaps. Tuned well, the results can be beautiful, but they are approximations of the physics.

Compared with ray tracing

• Rasterization: very fast, interactive, approximate light.
• Ray tracing and path tracing: slower, physically accurate light.
• Hybrid engines: rasterize the base image and use ray tracing for effects such as reflections and shadows.

In architectural visualization, many tools rely on rasterization, from SketchUp’s own viewport to real-time render engines. Knowing that their light is approximated helps explain why interiors in these engines often need extra fill lights.