Revisiting Photometric Ambiguity for Accurate Gaussian-Splatting Surface Reconstruction

Researchers introduced AmbiSuR, a Gaussian-splatting surface reconstruction framework that targets photometric ambiguity.

The method adds photometric disambiguation to better constrain geometry during differentiable rendering.

An ambiguity-indication module detects underconstrained regions and helps correct reconstruction errors.

Experiments show improved surface reconstruction performance across challenging scenes and ambiguity-heavy cases.