Research map / 3D vision / 3D generation
ReconViaGen: Towards Accurate Multi-view 3D Object Reconstruction via Generation
Jiahao Chang, Chongjie Ye, Yushuang Wu, Yuantao Chen, Yidan Zhang, Zhongjin Luo, Chenghong Li, Yihao Zhi, Xiaoguang Han
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 |
|---|---|---|---|
| add global geometry condition (VGGT features) to SS Flow (ablation) | TRELLIS-M · Dora-bench | PSNR: 16.706 -> 20.462 | Improved |
| add per-view condition (PVC) to SLAT Flow (ablation) | ReconViaGen with GGC · Dora-bench | PSNR: 20.462 -> 21.045 | Improved |
| add PVC on top of GGC (ablation) | ReconViaGen · Dora-bench | F-score: 0.941 -> 0.937 | No clear effect |
| add rendering-aware velocity compensation at inference (ablation) | ReconViaGen · Dora-bench | PSNR: 21.045 -> 22.632 | Improved |
| use global geometry condition (Condition Net tokens) for SS Flow instead of feature volume (ablation) | SS Flow · Dora-bench | PSNR: 16.229 -> 20.462 | Improved |
| use GGC instead of per-view tokens concatenation for SS Flow (ablation) | SS Flow · Dora-bench | PSNR: 19.749 -> 20.462 | Improved |
| use per-view condition instead of global condition for SLAT Flow (ablation) | SLAT Flow · Dora-bench | PSNR: 17.784 -> 22.632 | Improved |
| increase input views from 2 to 4 (ablation) | ReconViaGen · Dora-bench | PSNR: 19.568 -> 22.632 | Improved |
| increase input views from 4 to 8 (ablation) | ReconViaGen · Dora-bench | PSNR: 22.632 -> 23.067 | Improved |
| ReconViaGen vs Object VGGT + 3DGS with more views (limited-view sampling) | ReconViaGen · Dora-bench 10 views limited | PSNR: 17.121 -> 21.866 | Improved |
| generation prior added to fine-tuned object VGGT for pose estimation | ReconViaGen · Dora-bench camera pose | RRE: 7.257 -> 7.925 | Mixed |
| fine-tune VGGT with LoRA on object data | VGGT · Dora-bench camera pose | RRE: 8.575 -> 7.257 | Improved |
Later papers that compared with this method
4 later accepted papers report ReconViaGen as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Free-Range Gaussians: Non-Grid-Aligned Generative 3D Gaussian Reconstruction (ECCV 2026 · better on 5 of 6 shared results)
- ROAR-3D: Routing Arbitrary Views for High-Fidelity 3D Generation (ECCV 2026 · better on 5 of 5 shared results)
- GO-Renderer: Generative Object Rendering with 3D-aware Controllable Video Diffusion Models (ECCV 2026 · better on 4 of 4 shared results)
- World Tracing: Pixel-Aligned Geometry Beyond the Visible (NeurIPS 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 2 of 2 shared results
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
- Decompositional Neural Scene Reconstruction with Generative Diffusion Prior (CVPR 2025 · similarity 0.91)
- Ouroboros3D: Image-to-3D Generation via 3D-aware Recursive Diffusion (CVPR 2025 · similarity 0.91)
- GenFusion: Closing the Loop between Reconstruction and Generation via Videos (CVPR 2025 · similarity 0.90)
- DreamComposer: Controllable 3D Object Generation via Multi-View Conditions (CVPR 2024 · similarity 0.89)
- MVGenMaster: Scaling Multi-View Generation from Any Image via 3D Priors Enhanced Diffusion Model (CVPR 2025 · similarity 0.89)
- MVD-Fusion: Single-view 3D via Depth-consistent Multi-view Generation (CVPR 2024 · similarity 0.89)
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.
