Latent Intrinsics Emerge from Training to Relight

TL;DR

This paper introduces a data-driven relighting method that uses latent variables to represent scene intrinsics and lighting, achieving state-of-the-art results without explicit geometry or prior albedo examples.

Contribution

The method models intrinsics and lighting as latent variables, enabling effective relighting and albedo recovery without explicit geometric or prior information.

Findings

Achieves state-of-the-art relighting performance on real scenes.

Recovers albedo from latent intrinsics without prior examples.

Latent representations effectively encode scene properties.

Abstract

Image relighting is the task of showing what a scene from a source image would look like if illuminated differently. Inverse graphics schemes recover an explicit representation of geometry and a set of chosen intrinsics, then relight with some form of renderer. However error control for inverse graphics is difficult, and inverse graphics methods can represent only the effects of the chosen intrinsics. This paper describes a relighting method that is entirely data-driven, where intrinsics and lighting are each represented as latent variables. Our approach produces SOTA relightings of real scenes, as measured by standard metrics. We show that albedo can be recovered from our latent intrinsics without using any example albedos, and that the albedos recovered are competitive with SOTA methods.