A method of creating and/or refining a scene model representing a generalized light field and a relightable matter field. Data used in reconstruction includes images, video, object models, and/or scene models. The generalized light field represents light flowing in the scene. The relightable matter field represents the light interaction properties and emissivity of matter occupying the scene. Scene data structures can be implemented using Neural Networks and Physics-based structures, among others. The light and matter fields are at least partially disentangled.
– Popular Definition from Wikipedia