Light Field¶

The light field is a function describing the amount of light traveling in every direction through every point in space. It is a fundamental concept in computational photography.

The plenoptic function \(P(\theta, \phi, \lambda, t, V_x, V_y, V_z)\) captures all possible views of a scene (position, direction, wavelength, time).
For static scenes, rays can be parameterized by two planes (4D representation): \(L(u, v, s, t)\).
Light field cameras (e.g., Lytro) capture the 4D light field in a single exposure using a microlens array, enabling post-capture refocusing and depth estimation.

Projector-Camera Systems¶

Projector-camera (pro-cam) systems combine a projector and camera to enable:

Scene geometry recovery via structured light
Radiometric compensation (adapting projected images to non-white/non-flat surfaces)
Augmented reality on arbitrary surfaces
Dual photography (reconstructing the projector’s view using Helmholtz reciprocity)

Coded Photography¶

Coded photography modifies camera parameters (aperture, exposure, sensor) during capture to encode additional scene information:

Coded aperture: Non-standard aperture patterns that enable depth estimation and deblurring from a single image
Flutter shutter: Temporally coded exposure that preserves high-frequency information in motion-blurred images, enabling deblurring
Coded illumination: Spatially or temporally modulated lighting to separate direct and indirect illumination