Rasterization

Rasterization is the process of converting vector graphics or geometric primitives into a raster image composed of discrete picture elements called pixels. It is the core operation in most real-time rendering pipelines, enabling smooth display on raster devices such as computer monitors and mobile screens.

In a typical 3D graphics pipeline, a scene is transformed by vertex programs into screen-space coordinates. After clipping and perspective division, a rasterizer determines which pixels are covered by each primitive, usually a triangle. For every covered pixel, a fragment (potential pixel) is produced with interpolated attributes—such as color, texture coordinates, and depth. A fragment shader then computes the final color and other per-fragment data. The resulting color is written to a frame buffer after depth testing, blending, and optional stencil operations.

Key characteristics of rasterization include its speed and compatibility with hardware acceleration. It relies on a depth (z-buffer) to resolve visibility and typically uses perspective-correct interpolation to preserve geometric accuracy across the primitive. To reduce jagged edges, various anti-aliasing techniques are employed, such as multisample anti-aliasing (MSAA) and post-processing methods like FXAA or TXAA.

Rasterization is widely used for real-time rendering in video games and interactive graphics due to its efficiency and predictability. It is often contrasted with ray tracing, which simulates light paths for higher realism but can be computationally heavier. In modern engines, hybrid approaches combine rasterization with selective ray tracing to balance performance and visual quality.