multisample

Multisample is a term used to describe the use of multiple samples to improve fidelity, accuracy, or realism in a system. The concept appears in several domains, most notably computer graphics and audio, but the general idea can apply to any measurement or reproduction process that benefits from averaging or layering multiple data points.

In computer graphics, multisample commonly refers to multisample anti-aliasing (MSAA). MSAA reduces jagged edges by taking several samples within each pixel and using the coverage information to blend the final color. Typical configurations include 2x, 4x, or 8x sampling. The technique requires per-sample data for color, depth, and stencil, enabling smoother edges on polygon boundaries without rendering an authoring-level high-resolution image. However, MSAA does not address all forms of aliasing, such as texture aliasing or shader aliasing, and may be complemented by other methods like supersampling, FXAA, or temporal anti-aliasing (TAA).

In audio, multisampling refers to using multiple sampled recordings to reproduce a sound more realistically. A multisampled instrument library contains many samples across different pitches and dynamic levels, mapped to keys or velocity layers. This allows smoother transitions in timbre and volume but demands greater storage and processing power.

In a broader sense, multisampling also enjoys use as a general approach to reduce measurement error by aggregating multiple samples, a strategy common across scientific and engineering disciplines.