{"id":31,"date":"2026-05-07T00:00:12","date_gmt":"2026-05-07T00:00:12","guid":{"rendered":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf17\/3dgs-view-extrapolation\/"},"modified":"2026-05-08T03:30:04","modified_gmt":"2026-05-08T03:30:04","slug":"3dgs-view-extrapolation","status":"publish","type":"page","link":"https:\/\/mscvprojects.ri.cmu.edu\/2026teamf17\/3dgs-view-extrapolation\/","title":{"rendered":"RefineGS: Generalizing 3D Gaussian Splatting for Out of Distribution Views"},"content":{"rendered":"\n

Srinath Ravi, Fernando De la Torre<\/strong><\/p>\n\n\n\n

\n
\n
\n
\"\"<\/figure>\n<\/div>\n\n\n\n
\n

3D Gaussian Splatting (3DGS) delivers impressive real-time scene reconstruction, yet it often struggles with “floaters” and blurred textures when viewed from angles outside the original training data. RefineGS<\/strong> solves this by introducing a post-hoc, image-space correction module that cleans up corrupted renders without altering the underlying 3D representation. By leveraging the CroCo v2<\/strong> backbone for geometric consistency and a Conditional Flow Matching (CFM)<\/strong> framework for efficient, deterministic refinement, RefineGS restores photometric accuracy to unseen views. Our framework provides a principled, high-performance solution for eliminating artifacts and ensuring structural integrity in modern splat-based pipelines.<\/p>\n<\/div>\n<\/div>\n<\/div>\n\n\n\n

\"\"<\/span>
\n
\n

RefineGS<\/strong><\/h1>\n\n\n\n

Instead of the computationally expensive task of re-optimizing Gaussian primitives, we propose a modular, image-space solution. RefineGS<\/strong> treats initial renders as “noisy” starting points and refines them using a learned transformer that understands 3D relationships.<\/p>\n\n\n\n

Co<\/strong>ntributions:<\/h3>\n\n\n\n