FixAnything: 3D-Consistent Rendering Refinement via Video Priors
Primary Source: https://github.com/kvuong2711/fix-anything
Author: Khiem Vuong, Deva Ramanan, Srinivas Kaza · Year: 2026 · Publisher: ECCV 2026 / CMU
Summary
An ECCV 2026 framework that harnesses pretrained video generative diffusion models to automatically refine, sharpen, and restore temporal and multiview consistency to noisy 3D Gaussian splats and NeRF renderings.
Key Points
- Uses video diffusion temporal attention as a prior to fix floaters, blurry textures, and geometric inconsistencies in 3D scene captures.
- Iterative distillation preserves novel-view camera geometry while enhancing photographic fidelity.
- Significantly outperforms 2D image diffusion refinement baselines in temporal stability.