Free Geometry: Refining 3D Reconstruction from Longer Versions of Itself

TL;DR

Free Geometry introduces a test-time self-evolution framework for 3D reconstruction models, improving accuracy by leveraging multi-view consistency without requiring 3D ground truth.

Contribution

It enables feed-forward 3D reconstruction models to self-calibrate at test time using self-supervision from multiple views, enhancing accuracy and robustness.

Findings

Improves camera pose accuracy by 3.73% on average.

Enhances point map prediction accuracy by 2.88%.

Requires less than 2 minutes for recalibration per dataset.

Abstract

Feed-forward 3D reconstruction models are efficient but rigid: once trained, they perform inference in a zero-shot manner and cannot adapt to the test scene. As a result, visually plausible reconstructions often contain errors, particularly under occlusions, specularities, and ambiguous cues. To address this, we introduce Free Geometry, a framework that enables feed-forward 3D reconstruction models to self-evolve at test time without any 3D ground truth. Our key insight is that, when the model receives more views, it produces more reliable and view-consistent reconstructions. Leveraging this property, given a testing sequence, we mask a subset of frames to construct a self-supervised task. Free Geometry enforces cross-view feature consistency between representations from full and partial observations, while maintaining the pairwise relations implied by the held-out frames. This self-supervision allows for fast recalibration via lightweight LoRA updates, taking less than 2 minutes per dataset on a single GPU. Our approach consistently improves state-of-the-art foundation models, including Depth Anything 3 and VGGT, across 4 benchmark datasets, yielding an average improvement of 3.73% in camera pose accuracy and 2.88% in point map prediction. Code is available at https://github.com/hiteacherIamhumble/Free-Geometry .