Research map / 3D vision / Depth and stereo
Creat3r: Confidence Reaggregation for Exploration-aware Active 3D Reconstruction
Chih-Jung Tsai, Hwann-Tzong Chen, Tyng-Luh Liu
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 |
|---|---|---|---|
| replace COLMAP-subsampled init with Creat3r-MASt3R points | FisherRF · Mip-NeRF 360 (10 views) | PSNR: 12.625 -> 14.196 | Improved |
| replace COLMAP-subsampled init with Creat3r-MASt3R points | Lyu et al. · Mip-NeRF 360 (20 views) | PSNR: 15.446 -> 16.264 | Improved |
| replace COLMAP-subsampled init with Creat3r-MASt3R points | Kopanas & Drettakis · Mip-NeRF 360 (20 views) | PSNR: 15.658 -> 15.727 | Improved |
| use MASt3R matching instead of LightGlue | Creat3r · Mip-NeRF 360 (10 views) | PSNR: 16.040 -> 17.809 | Improved |
| use robust DLT point estimate instead of MASt3R predicted points (ablation) | Creat3r · Mip-NeRF 360 (9 scenes, 10 views) | PSNR: 16.479 -> 17.809 | Improved |
| add exploration term to confidence-only selection (ablation) | Creat3r · Mip-NeRF 360 (9 scenes, 10 views) | PSNR: 17.022 -> 17.809 | Improved |
| add confidence term to exploration-only selection (ablation) | Creat3r · Mip-NeRF 360 (9 scenes, 10 views) | LPIPS: 0.543 -> 0.523 | Improved |
| add confidence term to exploration-only selection (ablation) | Creat3r · Mip-NeRF 360 (9 scenes, 10 views) | SSIM: 0.502 -> 0.511 | Mixed |
| use Creat3r selection instead of FisherRF | FisherRF · Tanks&Temples F1 (%) | F1: 8.61 -> 18.05 | Improved |
| use Creat3r selection instead of Kopanas & Drettakis | Kopanas & Drettakis · Tanks&Temples recall (%) | recall: 0.58 -> 25.93 | Improved |
| use Creat3r with 20 views instead of 3DGS on all views | 3DGS · Tanks&Temples Truck | F1: 0.19 -> 0.19 | No clear effect |
| use Creat3r with 20 views instead of 3DGS on all views | 3DGS · Tanks&Temples Ignatius | F1: 0.04 -> 0.25 | Improved |
Closest papers by meaning
- Sparse-view Pose Estimation and Reconstruction via Analysis by Generative Synthesis (NeurIPS 2024 · similarity 0.87)
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis (ICML 2025 · similarity 0.86)
- Pow3R: Empowering Unconstrained 3D Reconstruction with Camera and Scene Priors (CVPR 2025 · similarity 0.86)
- BAT3R: Robust Online 3D Reconstruction with Bayesian Adaptive State Updates (NeurIPS 2026 · similarity 0.86)
- MV-DUSt3R+: Single-Stage Scene Reconstruction from Sparse Views In 2 Seconds (CVPR 2025 · similarity 0.86)
- GenNBV: Generalizable Next-Best-View Policy for Active 3D Reconstruction (CVPR 2024 · similarity 0.86)
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.
