Rasters use a data model in which data, usually images or continuous datasets, can be stored and represented visually as values within a grid of cells. Raster grid cells are assigned values that represent specific properties; these grid cell values usually can be decoded as colors. Consequently, a raster dataset is rather like a sheet of graph paper in which each cell contains a color that corresponds with the data represented by the cell. The resolution of a raster dataset is the number of cells on the X and Y axes of the raster grid.

Raster datasets with which most users will be familiar are digital images, including JPEG, TIFF, or GIF images. The individual raster grid cells of these images are referred to as pixels. If the resolution of such images is large enough, individual pixels will be difficult to discern with the naked eye (depending on the scale at which the image is viewed).

The amount of color data that can be stored in a given pixel depends on the format of the raster image. Common raster formats are JPEG, TIFF, GIF, PNG, and BMP.

Raster images are optimal primarily for the storage and display of continuous data sets, which model phenomena without distinct boundaries such as temperature gradients over a given area. Continuous data sets are difficult to display as vectors. A disadvantage of rasters is that raster image files can be very large depending on the resolution, color depth, and compression of the image.

Compare: vectors.