EvenSplat separates lighting effects from 3D scene appearance
EvenSplat: Coupled 2D-3D Decomposition for Gaussian Splatting under Exposure and Illumination Variation
Computer Vision and Pattern Recognition
Summary
When you take pictures of an object under different lighting or exposure conditions, the object can look very different depending on shadows and bright spots. Many 3D reconstruction methods confuse these lighting changes with the object's true shape and color. The authors present EvenSplat, a system that separates lighting effects from the actual 3D scene properties by linking 2D image lighting information with 3D geometric models. This method helps produce more accurate 3D reconstructions, especially in challenging lighting conditions.
What this means in practice
- •For 3d graphics developers: Improve 3D scene capture tools to handle uneven lighting and exposure differences during photogrammetry and model reconstruction.
- •For augmented reality developers: Generate more realistic 3D models that maintain consistent appearance under varying illumination for AR applications.
Authors
Tongyu Wu, Jacob Edwards, Ziteng Cui, Caigui Jiang, Cheng Wang
Abstract
A surface photographed under even light presents nearly the same appearance from every angle; the same surface under uneven light does not. Exposure changes between views, illumination varies within a single image, and locally strong light sources leave one region bright and its neighbor in shadow. Multi-view reconstruction methods such as 3D Gaussian Splatting treat these lighting artifacts as if they were properties of the scene, entangling capture-specific illumination with the geometry and color they recover. We present EvenSplat, a framework that separates the two. EvenSplat couples an image-space illumination decomposition with an illumination field carried by the Gaussians, so that the same explanation of the lighting is shared between the two-dimensional and three-dimensional views of the scene; a camera-response network and a local exposure-compensation module absorb the global and residual differences that remain across training images. Through extensive experiments across multiple datasets and diverse forms of uneven illumination (cross-view exposure, spatial illumination variation, and high-contrast lighting) on both real-world captured and simulated benchmarks, EvenSplat generally outperforms state-of-the-art methods, particularly under high-contrast illumination.