# CVPR 2026 3D Reconstruction Strategic Processing Brief

Source: https://gisbi-kim.github.io/cvpr2026-explorer/output/cvpr2026_explorer.html
Total explorer papers parsed: 4070
Broad candidate papers selected: 1417
Strict blog-reading candidate papers selected: 864
Seed papers for first-pass blog reading: 178

## Working Thesis

CVPR 2026 3D reconstruction is not one bucket. It is splitting into feed-forward geometry foundation models, Gaussian/radiance-field representations, pose-calibration-localization reliability, dynamic/4D scenes, and 3D generation/editing bridges. VGGT is useful as a lens, but too narrow as the only search term.

## Cluster Counts

- General 3D reconstruction / multiview: 592
- Mesh / surface / implicit / occupancy: 561
- Dataset / benchmark / evaluation: 425
- Gaussian / radiance field / view synthesis: 333
- Dynamic / 4D reconstruction: 271
- Depth / stereo / dense correspondence: 267
- Pose / calibration / localization: 256
- Mapping / autonomous / embodied: 251
- 3D generation / editing bridge: 151
- VGGT / feed-forward geometry: 67

## Editorial Triage

- core trend paper: 417
- important bridge paper: 215
- specialized geometry paper: 118
- adjacent / inspect manually: 114

## Blog Reading Buckets

- C. cluster representative: 284
- D. adjacent but useful context: 268
- A. thesis anchor: representation shift: 159
- A. thesis anchor: VGGT/feed-forward geometry: 67
- A. thesis anchor: dynamic/4D recon: 40
- B. bridge: reconstruction becomes mapping/world model: 33
- B. bridge: representation meets metric pose: 13

## Strong Cross-Cluster Signals

- General 3D reconstruction / multiview + Mesh / surface / implicit / occupancy: 455
- Dataset / benchmark / evaluation + General 3D reconstruction / multiview: 285
- Dataset / benchmark / evaluation + Mesh / surface / implicit / occupancy: 269
- Gaussian / radiance field / view synthesis + General 3D reconstruction / multiview: 228
- Dynamic / 4D reconstruction + General 3D reconstruction / multiview: 202
- Gaussian / radiance field / view synthesis + Mesh / surface / implicit / occupancy: 185
- Depth / stereo / dense correspondence + General 3D reconstruction / multiview: 173
- Depth / stereo / dense correspondence + Mesh / surface / implicit / occupancy: 171
- Dynamic / 4D reconstruction + Mesh / surface / implicit / occupancy: 171
- Dataset / benchmark / evaluation + Mapping / autonomous / embodied: 161
- Mapping / autonomous / embodied + Mesh / surface / implicit / occupancy: 160
- Dataset / benchmark / evaluation + Gaussian / radiance field / view synthesis: 148

## Cluster Reading Lists

### VGGT / feed-forward geometry
- [100] AVGGT: Rethinking Global Attention for Accelerating VGGT | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] CineScene: Implicit 3D as Effective Scene Representation for Cinematic Video Generation | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] Deformable Gaussian Occupancy: Decoupling Rigid and Nonrigid Motion with Factorized Distillation | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] Dynamic Visual SLAM using a General 3D Prior | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] DynamicVGGT: Learning Dynamic Point Maps for 4D Scene Reconstruction in Autonomous Driving | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] E-RayZer: Self-supervised 3D Reconstruction as Spatial Visual Pre-training | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] Emergent Extreme-View Geometry in 3D Foundation Models | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] Emergent Outlier View Rejection in Visual Geometry Grounded Transformers | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] FRM: Linear-Time 3D Reconstruction via Test-Time Training | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] FlashVGGT: Efficient and Scalable Visual Geometry Transformers with Compressed Descriptor Attention | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.

### General 3D reconstruction / multiview
- [100] $L^{2}DGS$: Low-Light Dynamic Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] $\textit{4DSurf}$: High-Fidelity Dynamic Scene Surface Reconstruction | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting from unposed Spike Stream | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Object Perception Transformer (3PT) | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4C4D: 4 Camera 4D Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4D Primitive-Mâché: Glueing Primitives for Persistent 4D Scene Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.

### Gaussian / radiance field / view synthesis
- [100] $L^{2}DGS$: Low-Light Dynamic Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] $\textit{4DSurf}$: High-Fidelity Dynamic Scene Surface Reconstruction | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting from unposed Spike Stream | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4C4D: 4 Camera 4D Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4DEquine: Disentangling Motion and Appearance for 4D Equine Reconstruction from Monocular Video | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] AERGS-SLAM: Auto-Exposure-Robust Stereo 3D Gaussian Splatting SLAM | core trend paper | Use as a bridge between reconstruction representation and metric pose/calibration reliability.
- [100] ActivePolicy: Active Gaussian Reconstruction and Optimization Strategy Based on Global-Local Information Gain | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.

### Pose / calibration / localization
- [100] 3D Gaussian Splatting from unposed Spike Stream | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Object Perception Transformer (3PT) | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] A Cross-view Fusion Framework for Robust 6-DoF Grasp Pose Estimation | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] AERGS-SLAM: Auto-Exposure-Robust Stereo 3D Gaussian Splatting SLAM | core trend paper | Use as a bridge between reconstruction representation and metric pose/calibration reliability.
- [100] AMB3R: Accurate Feed-forward Metric-scale 3D Reconstruction with Backend | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] AeroGS: Scale-Aware Gaussian Splatting for Pose-Free Dynamic UAV Scene Reconstruction | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignment | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] Breaking the 3D Dataset Bottleneck: Fast Scalable Generation of Aligned 3D Assets from Scratch  for Category 6D Pose Estimation and Robotic Grasping | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] CARD: A Multi-Modal Automotive Dataset for Dense 3D Reconstruction in Challenging Road Topography | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.

### Depth / stereo / dense correspondence
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Object Perception Transformer (3PT) | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D Primitive-Mâché: Glueing Primitives for Persistent 4D Scene Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] AERGS-SLAM: Auto-Exposure-Robust Stereo 3D Gaussian Splatting SLAM | core trend paper | Use as a bridge between reconstruction representation and metric pose/calibration reliability.
- [100] AMB3R: Accurate Feed-forward Metric-scale 3D Reconstruction with Backend | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] AVGGT: Rethinking Global Attention for Accelerating VGGT | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] ActivePolicy: Active Gaussian Reconstruction and Optimization Strategy Based on Global-Local Information Gain | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.

### Dynamic / 4D reconstruction
- [100] $L^{2}DGS$: Low-Light Dynamic Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] $\textit{4DSurf}$: High-Fidelity Dynamic Scene Surface Reconstruction | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting from unposed Spike Stream | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4C4D: 4 Camera 4D Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4D Primitive-Mâché: Glueing Primitives for Persistent 4D Scene Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4DEquine: Disentangling Motion and Appearance for 4D Equine Reconstruction from Monocular Video | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] ActivePolicy: Active Gaussian Reconstruction and Optimization Strategy Based on Global-Local Information Gain | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.

### Mesh / surface / implicit / occupancy
- [100] $\textit{4DSurf}$: High-Fidelity Dynamic Scene Surface Reconstruction | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Object Perception Transformer (3PT) | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D Primitive-Mâché: Glueing Primitives for Persistent 4D Scene Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4DEquine: Disentangling Motion and Appearance for 4D Equine Reconstruction from Monocular Video | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.

### Mapping / autonomous / embodied
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] A Cross-view Fusion Framework for Robust 6-DoF Grasp Pose Estimation | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] AERGS-SLAM: Auto-Exposure-Robust Stereo 3D Gaussian Splatting SLAM | core trend paper | Use as a bridge between reconstruction representation and metric pose/calibration reliability.
- [100] Breaking the 3D Dataset Bottleneck: Fast Scalable Generation of Aligned 3D Assets from Scratch  for Category 6D Pose Estimation and Robotic Grasping | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] CARD: A Multi-Modal Automotive Dataset for Dense 3D Reconstruction in Challenging Road Topography | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] Catch Me if You Can: Active Mapping of Moving 3D Objects | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] ChronoGS: Disentangling Invariants and Changes in Multi-Period Scenes | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] Clone Deterministic 3D Worlds | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Images | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] DLWM: Dual Latent World Models enable Holistic Gaussian-centric Pre-training in Autonomous Driving | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.

### 3D generation / editing bridge
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] BiMotion: B-spline Motion for Text-guided Dynamic 3D Character Generation | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] Breaking the 3D Dataset Bottleneck: Fast Scalable Generation of Aligned 3D Assets from Scratch  for Category 6D Pose Estimation and Robotic Grasping | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] Captain Safari: A Real-time World Engine | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] Catalyst4D: High-Fidelity 3D-to-4D Scene Editing via Dynamic Propagation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] CineScene: Implicit 3D as Effective Scene Representation for Cinematic Video Generation | core trend paper | Use as evidence that multi-view geometry is becoming a feed-forward foundation-model problem.
- [100] Dexterous World Models | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] Diff4Splat: Repurposing Video Diffusion Models  for  Dynamic Scene Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.

### Dataset / benchmark / evaluation
- [100] $L^{2}DGS$: Low-Light Dynamic Gaussian Splatting | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3D-Object Perception Transformer (3PT) | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 3DReflecNet: A Large-Scale Dataset for 3D Reconstruction of Reflective, Transparent, and Low-Texture Objects | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4D-RGPT: Toward Region-level 4D Understanding via Perceptual Distillation | core trend paper | Read early; it likely changes the framing of the 3D reconstruction cluster.
- [100] 4DEquine: Disentangling Motion and Appearance for 4D Equine Reconstruction from Monocular Video | core trend paper | Use as evidence that Gaussian/radiance representations are moving from static NVS toward dynamic scene models.
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.
- [100] A Cross-view Fusion Framework for Robust 6-DoF Grasp Pose Estimation | core trend paper | Use for the robotics/SLAM angle: reconstruction becomes a map or world model, not just a visual asset.

## First-Pass Seed List by Editorial Bucket

### A. thesis anchor: VGGT/feed-forward geometry
- [100] CineScene: Implicit 3D as Effective Scene Representation for Cinematic Video Generation | Generative Models / Video Generation
- [100] Deformable Gaussian Occupancy: Decoupling Rigid and Nonrigid Motion with Factorized Distillation | Autonomous Driving / Autonomous Driving
- [100] Dynamic Visual SLAM using a General 3D Prior | Robotics & Embodied AI / Embodied AI
- [100] DynamicVGGT: Learning Dynamic Point Maps for 4D Scene Reconstruction in Autonomous Driving | Autonomous Driving / Autonomous Driving
- [100] E-RayZer: Self-supervised 3D Reconstruction as Spatial Visual Pre-training | Learning Algorithms / Self-supervised
- [100] Emergent Extreme-View Geometry in 3D Foundation Models | 3D Vision & Geometry / 3D Reconstruction
- [100] Flow3r: Factored Flow Prediction for Visual Geometry Learning | 3D Vision & Geometry / 3D Reconstruction
- [100] FRM: Linear-Time 3D Reconstruction via Test-Time Training | 3D Vision & Geometry / 3D Reconstruction
- [100] GaussianGrow: Geometry-aware Gaussian Growing from 3D Point Clouds with Text Guidance | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Gen3R: 3D Scene Generation Meets Feed-Forward Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] Generalizable Sparse-View 3D Reconstruction from Unconstrained Images | 3D Vision & Geometry / 3D Reconstruction
- [100] Generalizing Visual Geometry Priors to Sparse Gaussian Occupancy Prediction | Autonomous Driving / Autonomous Driving
- [100] GGPT: Geometry-Grounded Point Transformer | 3D Vision & Geometry / 3D Reconstruction
- [100] Group Editing: Edit Multiple Images in One Go | Generative Models / Image Editing
- [100] HTTM: Head-wise Temporal Token Merging for Faster VGGT | 3D Vision & Geometry / 3D Reconstruction
- [100] LASER: Layer-wise Scale Alignment for Training-Free Streaming 4D Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] Learning 3D Reconstruction with Priors in Test Time | 3D Vision & Geometry / 3D Reconstruction
- [100] Learning to Drive is a Free Gift: Large-Scale Label-Free Autonomy Pretraining from Unposed In-The-Wild Videos | Autonomous Driving / Autonomous Driving
- [100] MoRE: 3D Visual Geometry Reconstruction Meets Mixture-of-Experts | 3D Vision & Geometry / 3D Reconstruction
- [100] MotionCrafter: Repurposing Video Generators for Dense Geometry and Motion Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] OmniVGGT: Omni-Modality Driven Visual Geometry Grounded Transformer | 3D Vision & Geometry / 3D Reconstruction
- [100] Pano3DComposer: Feed-Forward Compositional 3D Scene Generation from Single Panoramic Image | 3D Vision & Geometry / 3D Reconstruction
- [100] Point4Cast: Streaming Dynamic Scene Reconstruction and Forecasting | 3D Vision & Geometry / 3D Reconstruction
- [100] Points-to-3D: Structure-Aware 3D Generation with Point Cloud Priors | 3D Vision & Geometry / 3D Reconstruction
- [100] QVGGT: Post-Training Quantized Visual Geometry Grounded Transformer | 3D Vision & Geometry / 3D Reconstruction
- [100] Reliev3R: Relieving Feed-forward 3D Reconstruction from Multi-View Geometric Annotations | 3D Vision & Geometry / 3D Reconstruction
- [100] Scal3R: Scalable Test-Time Training for Feed-forward Large-Scale 3D Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] Unlocking the Power of Critical Factors for 3D Visual Geometry Estimation | Autonomous Driving / Autonomous Driving
- [100] VGGT-Det: Mining VGGT Internal Priors for Sensor-Geometry-Free Multi-view Indoor 3D Object Detection | 3D Vision & Geometry / 3D Reconstruction
- [100] VGGT-Segmentor: Geometry-Enhanced Cross-View Segmentation | Segmentation & Dense Prediction / Segmentation

### A. thesis anchor: dynamic/4D recon
- [100] 4D Primitive-Mâché: Glueing Primitives for Persistent 4D Scene Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] ArtHOI: Taming Foundation Models for Monocular 4D Reconstruction of Hand-Articulated-Object Interactions | 3D Vision & Geometry / 3D Reconstruction
- [100] Captain Safari: A Real-time World Engine | 3D Vision & Geometry / 3D Reconstruction
- [100] CARI4D: Category Agnostic 4D Reconstruction of Human-Object Interaction | 3D Vision & Geometry / 3D Reconstruction
- [100] Catch Me if You Can: Active Mapping of Moving 3D Objects | 3D Vision & Geometry / 3D Reconstruction
- [100] Clone Deterministic 3D Worlds | 3D Vision & Geometry / 3D Reconstruction
- [100] Complet4R: Geometric Complete 4D Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] Efficiently Reconstructing Dynamic Scenes one D4RT at a Time | 3D Vision & Geometry / 3D Reconstruction
- [100] EfficientMonoHair: Fast Strand-Level Reconstruction from Monocular Video via Multi-View Direction Fusion | 3D Vision & Geometry / 3D Reconstruction
- [100] ESAM++: Efficient Online 3D Perception on the Edge | 3D Vision & Geometry / 3D Reconstruction
- [100] Fast3Dcache: Training-free 3D Geometry Synthesis Acceleration | 3D Vision & Geometry / 3D Reconstruction
- [100] Featurising Pixels from Dynamic 3D Scenes with Linear In-Context Learners | 3D Vision & Geometry / 3D Reconstruction
- [100] FISHuman: Fine-grained Single-image 3D Human Reconstruction via Multi-view 4D Remeshing | 3D Vision & Geometry / 3D Reconstruction
- [100] GeoWorld: Geometric World Models | 3D Vision & Geometry / 3D Reconstruction
- [100] Inferring Compositional 4D Scenes without Ever Seeing One | 3D Vision & Geometry / 3D Reconstruction
- [100] MoRe: Motion-aware Feed-forward 4D Reconstruction Transrformer | 3D Vision & Geometry / 3D Reconstruction
- [100] Neu-PiG: Neural Preconditioned Grids for Fast Dynamic Surface Reconstruction on Long Sequences | 3D Vision & Geometry / 3D Reconstruction
- [100] NeuROK: Generative 4D Neural Object Kinematics | 3D Vision & Geometry / 3D Reconstruction
- [100] Order Matters: 3D Shape Generation from Sequential VR Sketches | 3D Vision & Geometry / 3D Reconstruction
- [100] PatchScene: Patch-based Voxel Diffusion Model for Large-Scale Scene Completion | 3D Vision & Geometry / 3D Reconstruction
- [100] ReFlow: Self-correction Motion Learning for Dynamic Scene Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] ReGenHOI: Unifying Reconstruction and Generation for 3D Human–Object Interaction Understanding | 3D Vision & Geometry / 3D Reconstruction
- [100] Revisiting Monocular SLAM with Spatio-Temporal Scene Modeling | 3D Vision & Geometry / Pose Estimation
- [100] Spatia: Video Generation with Updatable Spatial Memory | 3D Vision & Geometry / 3D Reconstruction
- [100] StereoWorld: Geometry-Aware Monocular-to-Stereo Video Generation | 3D Vision & Geometry / 3D Reconstruction
- [100] Tokenizing Vector Animation for Autoregresive Generation | 3D Vision & Geometry / 3D Reconstruction
- [100] Towards Visual Query Localization in the 3D World | 3D Vision & Geometry / 3D Reconstruction
- [100] VABench: A Comprehensive Benchmark for Audio-Video Generation | 3D Vision & Geometry / 3D Reconstruction
- [100] Vista4D: Video Reshooting with 4D Point Clouds | 3D Vision & Geometry / 3D Reconstruction
- [100] WorldStereo: Bridging Controllable Video Generation and Scene Reconstruction via 3D Geometric Memories | 3D Vision & Geometry / 3D Reconstruction

### A. thesis anchor: representation shift
- [100] 3D Gaussian Splatting with Self-Constrained Prior for High Fidelity Surface Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] 4DEquine: Disentangling Motion and Appearance for 4D Equine Reconstruction from Monocular Video | 3D Vision & Geometry / 3D Reconstruction
- [100] ActivePolicy: Active Gaussian Reconstruction and Optimization Strategy Based on Global-Local Information Gain | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] AeroGS: Scale-Aware Gaussian Splatting for Pose-Free Dynamic UAV Scene Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] AnchorSplat: Feed-Forward 3D Gaussian Splatting With 3D Geometric Priors | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] BA-GS: Bayesian Adaptive Gaussian Splatting for SFM-Free 3D Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] ClipGStream: Clip-Stream Gaussian Splatting for Any Length and Any Motion Multi-View Dynamic Scene Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] D-Prism: Differentiable Primitives for Structured Dynamic Modeling | 3D Vision & Geometry / 3D Reconstruction
- [100] Dark3R: Learning Structure from Motion in the Dark | 3D Vision & Geometry / 3D Reconstruction
- [100] Disco-GS: Gaussian Splatting in Dynamic Color Lighting | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] E2EGS: Event-to-Edge Gaussian Splatting for Pose-Free 3D Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Energy-GS: Image Energy-guided Pose Alignment Gaussian Splatting with redesigned pose gradient flow | 3D Vision & Geometry / Pose Estimation
- [100] FastEventDGS: Deformable Gaussian Splatting for Fast Dynamic Scenes from a Single Event Camera | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] FastGS: Training 3D Gaussian Splatting in 100 Seconds | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Feed-forward Gaussian Registration for Head Avatar Creation and Editing | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] GaussFusion: Improving 3D Reconstruction in the Wild with Geometry-Informed Video Generator | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Gaussian Mapping for Evolving Scenes | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] GeoDiff4D: Geometry-Aware Diffusion for 4D Head Avatar Reconstruction | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Geometric-Photometric Event-based 3D Gaussian Ray Tracing | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Learning 3D Representations for Spatial Intelligence from Unposed Multi-View Images | 3D Vision & Geometry / 3D Reconstruction
- [100] No Calibration, No Depth, No Problem: Cross-Sensor View Synthesis with 3D Consistency | 3D Vision & Geometry / 3D Reconstruction
- [100] Perceptual 3D Simulation With Physical World Modeling | 3D Vision & Geometry / 3D Reconstruction
- [100] Photo3D: Advancing Photorealistic 3D Generation through Structure‑Aligned Detail Enhancement | 3D Vision & Geometry / 3D Reconstruction
- [100] Pose-Free Omnidirectional Gaussian Splatting for 360-Degree Videos with Consistent Depth Priors | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] REArtGS++: Generalizable Articulation Reconstruction with Temporal Geometry Constraint via Planar Gaussian Splatting | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] RF4D:Neural Radar Fields for Novel View Synthesis in Outdoor Dynamic Scenes | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] RHINO: Reconstructing Human Interactions with Novel Objects from Monocular Videos | 3D Vision & Geometry / 3D Reconstruction
- [100] Scaling4D: Pushing the Frontier of Video Novel View Synthesis through Large-Scale Monocular Videos | 3D Vision & Geometry / 3D Reconstruction
- [100] Towards Realistic and Consistent Orbital Video Generation via 3D Foundation Priors | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Tracking-Guided 4D Generation: Foundation-Tracker Motion Priors for 3D Model Animation | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Unblur-SLAM: Dense Neural SLAM for Blurry Inputs | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] VarSplat: Uncertainty-aware 3D Gaussian Splatting for Robust RGB-D SLAM | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] VDFE: Difference-Aware 3D Scene Editing with Non-Intrusive Video Diffusion Priors for Multi-View Consistency and Efficiency | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] VerseCrafter: Dynamic Realistic Video World Model with 4D Geometric Control | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Zero-Shot Reconstruction of Animatable 3D Avatars with Cloth Dynamics from a Single Image | 3D Vision & Geometry / 3D Gaussian Splatting

### B. bridge: reconstruction becomes mapping/world model
- [100] ChronoGS: Disentangling Invariants and Changes in Multi-Period Scenes | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] DROID-SLAM in the Wild | 3D Vision & Geometry / Pose Estimation
- [100] Dual-Agent Reinforcement Learning for Adaptive and Cost-Aware Visual–Inertial Odometry | 3D Vision & Geometry / Pose Estimation
- [100] ForeHOI: Feed-forward 3D Object Reconstruction from Daily Hand-Object Interaction Videos | 3D Vision & Geometry / 3D Reconstruction
- [100] Haptic Neural Fields: Bringing Tactile Interactions to 3D Rendered Scenes | 3D Vision & Geometry / 3D Reconstruction
- [100] HOLO: Homography-Guided Pose Estimator Network for Fine-Grained Visual Localization on SD Maps | 3D Vision & Geometry / Pose Estimation
- [100] MajutsuCity: Language-driven Aesthetic-adaptive City Generation with Controllable 3D Assets and Layouts | 3D Vision & Geometry / 3D Reconstruction
- [100] PhysX-Anything: Simulation-Ready Physical 3D Assets from Single Image | 3D Vision & Geometry / 3D Reconstruction
- [100] PromptDepth: Efficient and Promptable Geometric 3D Vision Model \\ for Embodied Intelligence | 3D Vision & Geometry / 3D Reconstruction
- [100] Representing 3D Faces with Learnable B-Spline Volumes | 3D Vision & Geometry / 3D Reconstruction
- [100] SAGE: Scalable Agentic 3D Scene Generation for Embodied AI | 3D Vision & Geometry / 3D Reconstruction
- [100] SCE-SLAM: Scale-Consistent Monocular SLAM via Scene Coordinate Embeddings | 3D Vision & Geometry / Pose Estimation
- [100] SE(3)-Equivariance with Geometric and Topological Guidance for Category-Level Object Pose Estimation | 3D Vision & Geometry / Pose Estimation
- [100] SMVRT: Implicit Human 3D Modeling Using Sparse Multi-view Volumetric Reconstruction with Transformer Fusion | 3D Vision & Geometry / 3D Reconstruction
- [100] StreamVLO: Streaming Visual–LiDAR Odometry with Cumulative Drift Compensation | 3D Vision & Geometry / Pose Estimation
- [100] VLM-Loc: Localization in Point Cloud Maps via Vision-Language Models | 3D Vision & Geometry / Pose Estimation
- [100] Volumetric Functional Maps | 3D Vision & Geometry / 3D Reconstruction
- [99] Deep Feature Deformation Weights | 3D Vision & Geometry / 3D Reconstruction
- [99] HOG-Layout: Hierarchical 3D Scene Generation, Optimization and Editing via Vision-Language Models | 3D Vision & Geometry / 3D Reconstruction
- [99] OnlinePG: Online Open-Vocabulary Panoptic Mapping with 3D Gaussian Splatting | 3D Vision & Geometry / 3D Gaussian Splatting
- [99] SwiftTailor: Efficient 3D Garment Generation with Geometry Image Representation | 3D Vision & Geometry / 3D Reconstruction
- [99] UniTEX: Universal High Fidelity Generative Texturing for 3D Shapes | 3D Vision & Geometry / 3D Reconstruction
- [93] Hg-I2P: Bridging Modalities for Generalizable Image-to-Point-Cloud Registration via Heterogeneous Graphs | 3D Vision & Geometry / Pose Estimation
- [92] Seeing through Light and Darkness: Sensor-Physics Grounded Deblurring HDR NeRF from Single-Exposure Images and Events | 3D Vision & Geometry / 3D Gaussian Splatting
- [91] Reconstructing Functional 3D Scenes from Egocentric Interaction Videos | 3D Vision & Geometry / 3D Reconstruction

### B. bridge: representation meets metric pose
- [100] AERGS-SLAM: Auto-Exposure-Robust Stereo 3D Gaussian Splatting SLAM | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] DMAligner: Enhancing Image Alignment via Diffusion Model Based View Synthesis | 3D Vision & Geometry / Pose Estimation
- [100] Flow4DGS-SLAM: Optical Flow-Guided 4D Gaussian Splatting SLAM | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] ODGS-SLAM: Omnidirectional Gaussian Splatting SLAM | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] SGAD-SLAM: Splatting Gaussians at Adjusted Depth for Better Radiance Fields in RGBD SLAM | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] ULF-Loc: Unbiased Landmark Feature for Robust Visual Localization with 3D Gaussian Splatting | 3D Vision & Geometry / Pose Estimation
- [96] Rethinking Pose Refinement in 3D Gaussian Splatting under Pose Prior and Geometric Uncertainty | 3D Vision & Geometry / 3D Gaussian Splatting
- [94] Hierarchical Visual Relocalization with Nearest View Synthesis from Feature Gaussian Splatting | 3D Vision & Geometry / 3D Gaussian Splatting
- [88] FMPose: 3D Pose Estimation via Flow Matching | 3D Vision & Geometry / Pose Estimation
- [88] Uncertainty-Aware Modality Fusion for Unaligned RGB-T Salient Object  Detection | 3D Vision & Geometry / Pose Estimation
- [87] GenSplat: Bridging the Generalization Gap in 3DGS Language Comprehension | 3D Vision & Geometry / 3D Gaussian Splatting
- [69] GS^2: Graph-based Spatial Distribution Optimization for Compact 3D Gaussian Splatting | 3D Vision & Geometry / 3D Gaussian Splatting
- [65] Landscape-Awareness for Geometric View Diffusion Model | 3D Vision & Geometry / Pose Estimation

### C. cluster representative
- [100] 3D-Fixer: Coarse-to-Fine In-place Completion for 3D Scenes  from a Single Image | 3D Vision & Geometry / 3D Reconstruction
- [100] 3D-Object Perception Transformer (3PT) | 3D Vision & Geometry / 3D Reconstruction
- [100] AlignPose: Generalizable 6D Pose Estimation via Multi-view Feature-metric Alignment | 3D Vision & Geometry / Pose Estimation
- [100] AMB3R: Accurate Feed-forward Metric-scale 3D Reconstruction with Backend | 3D Vision & Geometry / 3D Reconstruction
- [100] ArchSym: Detecting 3D-Grounded Architectural Symmetries in the Wild | 3D Vision & Geometry / 3D Reconstruction
- [100] CoLoR: The Devil is in Scene Coordinate Regression for Large-Scale Visual Localization | 3D Vision & Geometry / Pose Estimation
- [100] Cupid: Generative 3D Reconstruction via Joint Object and Pose Modeling | 3D Vision & Geometry / 3D Reconstruction
- [100] Diff4Splat: Repurposing Video Diffusion Models  for  Dynamic Scene Generation | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] Edges Compete for Trust: Group Relative Edge Optimization for Building Reconstruction from Point Clouds | 3D Vision & Geometry / 3D Reconstruction
- [100] Event6D: Event-based Novel Object 6D Pose Tracking | 3D Vision & Geometry / Pose Estimation
- [100] Extend3D: Town-scale 3D Generation | 3D Vision & Geometry / 3D Reconstruction
- [100] Fast Markov Random Field Optimisation for Topologically Noisy 3D Shape Matching | 3D Vision & Geometry / Pose Estimation
- [100] From Feature Learning to Spectral Basis Learning: A Unifying and Flexible Framework for Efficient and Robust Shape Matching | 3D Vision & Geometry / Pose Estimation
- [100] FunFact: Building Probabilistic Functional 3D Scene Graphs via Factor-Graph Reasoning | 3D Vision & Geometry / 3D Reconstruction
- [100] FUSER: Feed-Forward Multiview 3D Registration Transformer and SE(3)$^N$ Diffusion Refinement | 3D Vision & Geometry / Pose Estimation
- [100] JRM: Joint Reconstruction Model for Multiple Objects without Alignment | 3D Vision & Geometry / 3D Reconstruction
- [100] Long-Tail Internet Photo Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] ManifoldNeuS: Manifold-aware View Optimizability for Pose-Free Neural Surface Reconstruction | 3D Vision & Geometry / 3D Reconstruction
- [100] Native and Compact Structured Latents for 3D Generation | 3D Vision & Geometry / 3D Reconstruction
- [100] Neural Field-Based 3D Surface Reconstruction of Microstructures from Multi-Detector Signals in Scanning Electron Microscopy | 3D Vision & Geometry / 3D Reconstruction
- [100] Pano360: Perspective to Panoramic Vision with Geometric Consistency | 3D Vision & Geometry / 3D Reconstruction
- [100] PanoVGGT: Feed-Forward 3D Reconstruction from Panoramic Imagery | 3D Vision & Geometry / 3D Reconstruction
- [100] PiLoT: Neural Pixel-to-3D Registration for UAV-based Ego and Target Geo-localization | 3D Vision & Geometry / Pose Estimation
- [100] PoseGAM: Robust Unseen Object Pose Estimation via Geometry-Aware Multi-View Reasoning | 3D Vision & Geometry / Pose Estimation
- [100] Registration-Free Learnable Multi-View Capture of Faces in Dense Semantic Correspondence | 3D Vision & Geometry / 3D Reconstruction
- [100] Rethinking 2D-3D Registration: A Novel Network for High-Value Zone Selection and Representation Consistency Alignment | 3D Vision & Geometry / Pose Estimation
- [100] SceneMaker: Open-set 3D Scene Generation with Decoupled De-occlusion and Pose Estimation Model | 3D Vision & Geometry / 3D Reconstruction
- [100] ShapeR: Robust Conditional 3D Shape Generation from Casual Captures | 3D Vision & Geometry / 3D Reconstruction
- [100] SpatialVID: A Large-Scale Video Dataset with Spatial Annotations | 3D Vision & Geometry / Pose Estimation
- [100] TagSplat: Topology-Aware Gaussian Splatting for Dynamic Mesh Modeling and Tracking | 3D Vision & Geometry / 3D Gaussian Splatting
- [100] TopoMesh: High-Fidelity Mesh Autoencoding via Topological Unification | 3D Vision & Geometry / 3D Reconstruction
- [100] UniCorn: Unified Correspondence Transformer Across 2D and 3D | 3D Vision & Geometry / 3D Reconstruction
- [100] UniSH: Unifying Scene and Human Reconstruction in a Feed-Forward Pass | 3D Vision & Geometry / 3D Reconstruction
- [100] ViLearn: Accelerating Training Convergence of Image-to-3D Generation via Visibility Learning | 3D Vision & Geometry / 3D Reconstruction
- [100] Yo'City: Personalized and Boundless 3D Realistic City Scene Generation via Self-Critic Expansion | 3D Vision & Geometry / 3D Reconstruction

### D. adjacent but useful context
- [100] 3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation | Generative Models / Video Generation
- [100] 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models | Data & Evaluation / Benchmark
- [100] Breaking the 3D Dataset Bottleneck: Fast Scalable Generation of Aligned 3D Assets from Scratch  for Category 6D Pose Estimation and Robotic Grasping | Data & Evaluation / Benchmark
- [100] Dexterous World Models | Robotics & Embodied AI / Embodied AI
- [100] DGGT: Feedforward 4D Reconstruction of Dynamic Driving Scenes using Unposed Images | Autonomous Driving / Autonomous Driving
- [100] Ego-1K – A Large-Scale Multiview Video Dataset for Egocentric Vision | Data & Evaluation / Benchmark
- [100] FaithFusion: Harmonizing Reconstruction and Generation via Pixel-wise Information Gain | Autonomous Driving / Autonomous Driving
- [100] RAYNOVA: Geometry-Free Auto-Regressive 4D World Modeling with Unified Spatio-Temporal Representation | Generative Models / Video Generation
- [100] UniPixie: Unified and Probabilistic 3D Physics Learning via Flow Matching | Robotics & Embodied AI / Embodied AI
- [100] Video2Robo: 3DGS-based Synthetic Data from One Video Enables Scalable Robot Learning | Robotics & Embodied AI / Embodied AI

## Blog Structure Draft

1. Why VGGT is the wrong but useful keyword
2. Feed-forward geometry as the new center of 3D reconstruction
3. Gaussian/radiance fields move from pretty rendering to maps, editing, and dynamic scenes
4. Pose, calibration, and localization become the reliability layer
5. Dynamic/4D reconstruction is the pressure test
6. What is central, what is incremental, and what is mostly application wrapping
7. Research wedge: metric online geometry priors for SLAM/robotics

## Manual Review Protocol

- Read all `core trend paper` rows first; these shape the thesis.
- Then read `important bridge paper` rows that connect VGGT, Gaussian maps, dynamic scenes, and pose/calibration.
- Use `specialized geometry paper` rows for subsection examples, not for the opening thesis unless they introduce a new evaluation axis.
- Treat `adjacent / inspect manually` rows as false-positive candidates until abstract-level evidence proves otherwise.
- For each selected paper, write one sentence for `what changed in the field` and one sentence for `what remains brittle or narrow`.