Research map / 3D vision / Depth and stereo
WorldMirror: Universal 3D World Reconstruction with Any-Prior Prompting
Yifan Liu, Zhiyuan Min, Zhenwei Wang, Junta Wu, Tengfei Wang, Yixuan Yuan, Yawei Luo, Chunchao Guo
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use single token instead of dense Plucker embedding for poses (ablation) | WorldMirror · ETH3D+DTU (images and poses) | average: 60.44 -> 61.06 | Improved |
| use single token instead of dense raymap for intrinsics (ablation) | WorldMirror · ETH3D+DTU (images and intrinsics) | average: 66.58 -> 68.96 | Improved |
| add GT camera parameters for 3DGS rendering (ablation) | WorldMirror · RealEstate10K 2 views | PSNR: 20.30 -> 20.29 | No clear effect |
| add novel-view rendering supervision (ablation) | WorldMirror · RealEstate10K 2 views | PSNR: 18.51 -> 20.29 | Improved |
| GS head predicts positions itself (add separate GS DPT design) (ablation) | WorldMirror · VR-NeRF 32 views | PSNR: 25.08 -> 25.75 | Improved |
| decoupled sequential training instead of joint training (ablation) | WorldMirror · 7-Scenes | accuracy error: 0.048 -> 0.043 | Improved |
| decoupled sequential training instead of joint training (ablation) | WorldMirror · NRGBD | accuracy error: 0.084 -> 0.041 | Improved |
| decoupled sequential training instead of joint training (ablation) | WorldMirror · DTU | completion error: 1.792 -> 1.780 | Mixed |
| prior dropout 0.5 instead of 0.0 (ablation) | WorldMirror · 7-Scenes (no prior) | accuracy error: 0.068 -> 0.044 | Improved |
| prior dropout 0.5 instead of 0.0 (ablation) | WorldMirror · 7-Scenes (all priors) | accuracy error: 0.015 -> 0.018 | Hurt |
| add camera pose, intrinsics and depth priors together | WorldMirror · 7-Scenes point map | accuracy error: 0.043 -> 0.018 | Improved |
| add depth prior | WorldMirror · DTU point map | accuracy error: 1.017 -> 0.831 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 5 of 5 shared results
- DepthSplat — DepthSplat: Connecting Gaussian Splatting and Depth (CVPR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Fostering Generalization in Single-View 3D Reconstruction by Learning a Hierarchy of Local and Global Shape Priors (CVPR 2021 · similarity 0.87)
- Pow3R: Empowering Unconstrained 3D Reconstruction with Camera and Scene Priors (CVPR 2025 · similarity 0.86)
- Depth2Pose: A Pose-Based Benchmark for Monocular Depth Estimation without Ground-Truth Depth (NeurIPS 2026 · similarity 0.86)
- UniK3D: Universal Camera Monocular 3D Estimation (CVPR 2025 · similarity 0.85)
- MegaSynth: Scaling Up 3D Scene Reconstruction with Synthesized Data (CVPR 2025 · similarity 0.85)
- Fin3R: Fine-tuning Feed-forward 3D Reconstruction Models via Monocular Knowledge Distillation (NeurIPS 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
