US Patent Application 17849055. LEARNING NEURAL REFLECTANCE SHADERS FROM IMAGES simplified abstract
Contents
LEARNING NEURAL REFLECTANCE SHADERS FROM IMAGES
Organization Name
Inventor(s)
Benjamin Ummenhofer of Unterhaching (DE)
Shenlong Wang of Santa Clara CA (US)
Sanskar Agrawal of Santa Clara CA (US)
Yixing Lao of Santa Clara CA (US)
Kai Zhang of Santa Clara CA (US)
Stephan Richter of Neubiberg (DE)
Vladlen Koltun of Santa Clara CA (US)
LEARNING NEURAL REFLECTANCE SHADERS FROM IMAGES - A simplified explanation of the abstract
- This abstract for appeared for US patent application number 17849055 Titled 'LEARNING NEURAL REFLECTANCE SHADERS FROM IMAGES'
Simplified Explanation
This abstract describes a technique for using machine learning to learn how light interacts with objects in images. The technique involves training a set of machine learning models to optimize the lighting and surface properties of an object based on input images. A shader is then generated using these models, which allows for rendering a 3D representation of the object with realistic lighting and surface appearance.
Original Abstract Submitted
Described herein are techniques for learning neural reflectance shaders from images. A set of one or more machine learning models can be trained to optimize an illumination latent code and a set of reflectance latent codes for an object within a set of input images. A shader can then be generated based on a machine learning model of the one or more machine learning models. The shader is configured to sample the illumination latent code and the set of reflectance latent codes for the object. A 3D representation of the object can be rendered using the generated shader.
