Raster Graphics are images represented as a rectangular grid of discrete samples, commonly called pixels. Each pixel stores one or more numerical values describing properties such as colour, brightness, opacity or other image information. Raster representation is particularly effective for photographs, scanned documents, digital paintings, textures and other imagery in which colour can vary independently from one location to another.
The quality and detail of a raster image are strongly related to its spatial resolution, because enlarging an image cannot reveal information that was never sampled. Raster Graphics can use different colour models, channel depths and storage arrangements, and they can be stored using uncompressed or compressed formats. The term therefore describes an image representation rather than a specific file format, compression method or visual style.
The basic structure of Raster Graphics is a two-dimensional sampling grid. A raster image with a given width and height contains a corresponding number of sample positions, although the physical size represented by each pixel depends on the display or print resolution. Each pixel can contain channels for red, green and blue, for example, along with an optional alpha channel describing transparency.
Greater bit depth allows more numerical values per channel and can reduce visible quantisation in gradients while providing greater latitude for image processing. Some professional workflows use high-bit-depth or floating-point representations to preserve information during compositing and colour correction. Raster images can also contain metadata describing colour profiles, orientation, capture information and other properties. The numerical representation is separate from how the image is eventually displayed, because colour-management systems interpret the stored values according to the relevant colour space and transfer characteristics.
Raster Graphics are fundamentally different from vector representations because raster images explicitly sample visual information at discrete positions. A vector image instead describes geometric objects through mathematical definitions such as paths, curves and fills. When a vector image is displayed, those definitions are generally rasterised into pixels for the target device.
Raster graphics therefore have a fixed sampling structure for a particular image resolution, whereas vector descriptions can often be rendered at different resolutions without the same direct loss of geometric precision. This distinction does not mean raster images are inherently lower quality: photographs and complex natural imagery are usually much more efficiently represented as rasters because describing every local variation as geometry would be impractical. Conversely, logos, diagrams and typography can often benefit from vector representation because their underlying shapes are relatively compact and scalable.
Resolution is a central concept in Raster Graphics. Spatial resolution describes the number of samples available across an image, while pixel density describes how those samples are distributed over a physical display or printed area. Increasing resolution can preserve finer detail, but it also increases storage, processing and transmission requirements. Upscaling interpolates additional pixel values from existing samples but does not recreate genuine missing detail, while downscaling reduces the number of samples and can remove high-frequency information.
Resampling filters determine how these transformations behave and can introduce blur, ringing or aliasing depending on the method. Aliasing occurs when image features finer than the sampling grid are represented inadequately, producing effects such as jagged edges or moiré patterns. Anti-aliasing and appropriate sampling strategies mitigate these problems by accounting for subpixel information.
Colour representation is another defining aspect of Raster Graphics. RGB representations are common for displays, while other models and colour spaces serve printing, video, scientific imaging and specialised applications. Bit depth determines the number of possible values represented by each channel, but higher numerical precision does not automatically guarantee a wider colour gamut because gamut is determined by the colour space.
High-dynamic-range imagery may use floating-point values or specialised integer encodings to preserve information outside the range of conventional display-oriented images. Alpha channels allow pixels to carry opacity information, which is useful for compositing. Premultiplied alpha represents colour values already multiplied by opacity and can provide mathematically convenient behaviour during compositing, whereas straight alpha stores colour independently of opacity. Confusing these representations can create dark or light fringes around transparent objects.
Raster Graphics support extensive image-processing operations because every pixel can be manipulated numerically. Operations include cropping, scaling, rotation, sharpening, blurring, denoising, edge detection, colour correction, compositing, histogram modification and convolution-based filtering. Some operations can be performed efficiently using local neighbourhoods, while others require global information or sophisticated optimisation. Raster editing systems often provide layers and masks so that changes can be made without permanently altering the underlying pixels. Raster graphics are also the natural output of most cameras and scanners because those devices sample incoming light or physical documents across a grid. Consequently, raster processing forms a major part of photography, publishing, medical imaging, computer vision and digital art.
Common raster file formats differ in compression, colour support, transparency and intended use. Some formats are designed for lossless storage, preserving the exact decoded pixel values, while others use lossy compression to reduce file size by discarding information considered less perceptually important. Portable images intended for web delivery may prioritise compactness and browser support, whereas production formats may prioritise high bit depth, transparency, metadata or editing flexibility.
A file extension alone does not fully describe the image's properties because formats can contain different encoding options. Raster images can also be embedded inside other documents or used as textures in three-dimensional graphics. Thus, Raster Graphics are not synonymous with any particular file format, and a robust understanding requires distinguishing representation, encoding, compression and containerisation.
Raster Graphics are advantageous when representing highly detailed and continuously varying imagery, but their resolution dependence creates limitations for certain applications. Enlarging raster artwork can expose pixelation or blur, while very high-resolution imagery can require substantial storage and processing bandwidth. Lossy compression can introduce blocking, ringing, smearing or other artefacts, while excessive processing can amplify noise and reduce detail. Nevertheless, raster representation is exceptionally general because any visual field that can be sampled can be represented as pixels.
Modern displays, cameras, scanners, image editors, scientific instruments and computer-vision systems all rely extensively on raster data. The distinction between raster and vector is therefore primarily about representation and sampling rather than a simple distinction between digital and analogue, because both are digital graphical representations and can coexist within the same production workflow.
