A vector graphic stores geometric descriptions—such as paths, curves, fills and strokes—instead of storing a fixed color sample for every output pixel.

Because the geometry is represented independently of one pixel grid, it can be transformed and then rasterized at the resolution needed for a display, printer or image file. A circle can therefore remain mathematically smooth when enlarged rather than exposing a previously stored grid of pixels.

Vector representation is especially effective for diagrams, typography and artwork built from clean geometric shapes. Raster images are usually more natural for densely varying sampled data such as photographs.

The distinction is a representation trade-off, not a quality hierarchy. Complex vector scenes can be expensive to render, while a raster image is cheap to display once sampled. Conversely, storing many high-resolution raster versions can cost more space than storing compact geometry and rendering it when needed.

Displaying a vector graphic still ends in rasterization whenever the output device is pixel based.