FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors
Rendering views using 3D scene representations such as Gaussian Splatting (3DGS), Neural Radiance Fields (NeRF), meshes, or even point clouds produces artifacts when input views are sparse or target views lie far from the input. Recent work mitigates these artifacts using diffusion-based generative priors, but is specialized to individual representations and require custom architectures or extensive retraining. We present FixAnything, a single model for fixing a wide range of rendering artifacts. It does so by repurposing a pretrained video generative model, leveraging its implicit multi-view
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- FuzzySimilar title/name (fuzzy) · 59%nvidia/LocateAnything-3B →
“Fuzzy title match (0.73): “FixAnything: 3D-Consistent Rendering Refinement via Video Ge” ≈ “nvidia/LocateAnything-3B””
- FuzzySimilar title/name (fuzzy) · 84%GoogleCloudPlatform/generative-ai →
“Fuzzy title match (0.92): “FixAnything: 3D-Consistent Rendering Refinement via Video Ge” ≈ “GoogleCloudPlatform/generative-ai””
- FuzzySimilar title/name (fuzzy) · 84%Mintplex-Labs/anything-llm →
“Fuzzy title match (0.92): “FixAnything: 3D-Consistent Rendering Refinement via Video Ge” ≈ “Mintplex-Labs/anything-llm””
- FuzzySimilar title/name (fuzzy) · 59%Developer-Y/cs-video-courses →
“Fuzzy title match (0.73): “FixAnything: 3D-Consistent Rendering Refinement via Video Ge” ≈ “Developer-Y/cs-video-courses””
- FuzzySimilar title/name (fuzzy) · 59%steven2358/awesome-generative-ai →
“Fuzzy title match (0.73): “FixAnything: 3D-Consistent Rendering Refinement via Video Ge” ≈ “steven2358/awesome-generative-ai””
- LinkedLinked via arxiv author · 85%Khiem Vuong →
“FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors”
- LinkedLinked via arxiv author · 85%Deva Ramanan →
“FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors”
- LinkedLinked via arxiv author · 85%Srinivasa Narasimhan →
“FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors”