Intrinsic decomposition and editing of 3D Gaussian splats

2026-06-30

AI Summary

This work introduces intrinsic decomposition into 3D Gaussian splatting radiance fields for the first time, aiming to disentangle illumination and material properties to enable editing. The method constructs a hierarchical set of Gaussian primitives to separately model albedo, shading, and view-dependent residuals, and employs differentiable optimization leveraging multi-view images and data-driven priors. This framework allows users to edit surface albedo from a single image and synthesize relit renderings that maintain consistent illumination across arbitrary viewpoints. The key contribution lies in achieving effective intrinsic decomposition within the 3D Gaussian representation, establishing the first material editing pipeline that supports single-image guidance while ensuring multi-view consistency.

Abstract

Intrinsic decomposition which expresses image colors as the product of diffuse albedo and shading, possibly augmented with view-dependent residuals has a long history in image editing as it enables the modification of object colors and textures without altering lighting. We extend intrinsic decomposition to radiance fields represented with Gaussian splatting by proposing solutions to three key aspects of such decomposition. First, we describe how to model the intrinsic decomposition as independent sets of Gaussian primitives, which allows each set to adapt to the characteristics of the layer it represents. Second, we present an optimization procedure guided by data-driven predictions to disentangle multi-view photographs of a scene into the aforementioned intrinsic sets. Finally, we provide an editing workflow where users modify the texture of planar surfaces simply by modifying the albedo of that surface in one image. Capturing this edit within the intrinsic radiance field allows re-rendering of the edited scene with plausible lighting under arbitrary viewpoints.

Problem

• intrinsic decomposition
• 3D Gaussian splatting
• radiance fields
• image editing
• albedo-shading separation

Innovation

• intrinsic decomposition
• 3D Gaussian splatting
• radiance fields
• albedo editing
• multi-view disentanglement