Lightfield

A light field is a representation of the light in a scene that captures how much light travels through every point in space in every direction. In the most common formulation, the four-dimensional light field is written as L(x, y, θ, φ), describing the amount of light passing through spatial coordinates (x, y) with direction angles (θ, φ). This 4D model can be extended to higher dimensions by including wavelength or time, giving a more complete description of the radiance in a scene.

The concept separates radiometry from geometry, allowing post-capture operations that would otherwise require multiple images or explicit 3D modeling. In practice, a light field records both spatial and angular information, enabling synthetic refocusing, perspective changes, and depth estimation from a single capture. Representations can be stored as plenoptic data sets, often organized as a 2D array of views or a 4D function sampled by a microlens array or camera arrangement.

Capture methods include plenoptic cameras, which place a microlens array in front of an image sensor to encode angular information, and camera arrays that sample the scene from multiple viewpoints. The resulting data enables computational photography tasks such as refocusing after the shot, depth map creation, and 3D reconstruction, as well as applications in immersive displays and virtual reality.

Historically, the idea of a plenoptic function was introduced in the 1930s, with formal light-field frameworks developed in later decades. The concept gained practical visibility in the 2000s and was popularized for consumers by devices like light-field cameras introduced in the early 2010s.

Applications of light-field data span computational imaging, graphics rendering, and display technology. Common challenges include balancing spatial and angular resolution, managing large data sizes, and achieving robust calibration and noise handling for real-world scenes.