High-fidelity three-dimensional asset encoding

Abstract

Certain aspects and features of this disclosure relate to rendering images by training a neural material and applying the material map to a coarse geometry to provide high-fidelity asset encoding. For example, training can involve sampling for a set of lighting and camera configurations arranged to render an image of a target asset. A value for a loss function comparing the target asset with the neural material can be optimized to train the neural material to encode a high-fidelity model of the target asset. This technique restricts the application of the neural material to a specific predetermined geometry, resulting in a reproducible asset that can be used efficiently. Such an asset can be deployed, as examples, to mobile devices or to the web, where the computational budget is limited, and nevertheless produce highly detailed images.