Light field rendering is a technique for generating new views from arbitrary camera positions without depth information or feature matching, by combining and resampling available images. The key idea is to interpret input images as 2D slices of a 4D light field function, which characterizes the flow of light through unobstructed space in a static scene. A sampled representation of light fields allows efficient creation and display of inward and outward looking views. Light fields can be created from large arrays of rendered or digitized images. A compression system compresses light fields by more than 100:1 with minimal fidelity loss. Antialiasing during creation and resampling during slice extraction are addressed. The light field is represented as a 4D function on the space of oriented lines. A light slab representation is used, where lines are parameterized by their intersections with two planes. This representation allows efficient geometric calculations and supports random access. The light field is created from rendered or digitized images, with a focus on generating a 4D light field from a set of images. The creation process involves sampling and resampling, with prefiltering to avoid aliasing. A computer-controlled camera gantry is used to acquire digitized images, with careful control of lighting and camera motion to ensure a static light field. The light field is compressed using a two-stage pipeline: vector quantization followed by entropy coding. This allows high compression rates, with the light field compressed by over 100:1. The decompression process involves retrieving codebook indices and dequantizing the samples. The display process involves resampling the radiance at the appropriate line parameters, with interpolation techniques used to reduce aliasing artifacts. The results show that the light field technique can generate high-quality images with minimal artifacts, even when compressed. The technique is efficient and robust, allowing real-time interaction with 3D scenes. The method has several limitations, including the need for high sampling density, restrictions on the observer's position, and the requirement for fixed illumination. Future work includes extending the technique to handle interreflections and exploring new representations for light in computer graphics and computer vision.Light field rendering is a technique for generating new views from arbitrary camera positions without depth information or feature matching, by combining and resampling available images. The key idea is to interpret input images as 2D slices of a 4D light field function, which characterizes the flow of light through unobstructed space in a static scene. A sampled representation of light fields allows efficient creation and display of inward and outward looking views. Light fields can be created from large arrays of rendered or digitized images. A compression system compresses light fields by more than 100:1 with minimal fidelity loss. Antialiasing during creation and resampling during slice extraction are addressed. The light field is represented as a 4D function on the space of oriented lines. A light slab representation is used, where lines are parameterized by their intersections with two planes. This representation allows efficient geometric calculations and supports random access. The light field is created from rendered or digitized images, with a focus on generating a 4D light field from a set of images. The creation process involves sampling and resampling, with prefiltering to avoid aliasing. A computer-controlled camera gantry is used to acquire digitized images, with careful control of lighting and camera motion to ensure a static light field. The light field is compressed using a two-stage pipeline: vector quantization followed by entropy coding. This allows high compression rates, with the light field compressed by over 100:1. The decompression process involves retrieving codebook indices and dequantizing the samples. The display process involves resampling the radiance at the appropriate line parameters, with interpolation techniques used to reduce aliasing artifacts. The results show that the light field technique can generate high-quality images with minimal artifacts, even when compressed. The technique is efficient and robust, allowing real-time interaction with 3D scenes. The method has several limitations, including the need for high sampling density, restrictions on the observer's position, and the requirement for fixed illumination. Future work includes extending the technique to handle interreflections and exploring new representations for light in computer graphics and computer vision.