Research map / 3D vision / Face and shape
3D Mesh Editing using Masked LRMs
Will Gao, Dilin Wang, Yuchen Fan, Aljaz Bozic, Tuur Stuyck, Zhengqin Li, Zhao Dong, Rakesh Ranjan, Nikolaos Sarafianos
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 8 input views instead of 6 (ablation) | Masked LRM · ABO | PSNR: 28.37 -> 28.65 | Improved |
| use 8 input views instead of 6 (ablation) | Masked LRM · GSO | PSNR: 27.24 -> 27.58 | Improved |
| use 8 input views instead of 6 (ablation) | Masked LRM · ABO | LPIPS: 0.081 -> 0.078 | Improved |
| Masked LRM instead of MeshLRM | MeshLRM · ABO | PSNR: 26.09 -> 28.65 | Improved |
| Masked LRM instead of MeshLRM | MeshLRM · GSO | PSNR: 27.93 -> 27.58 | No clear effect |
| Masked LRM instead of InstantMesh (NeRF) | InstantMesh · GSO | PSNR: 23.14 -> 27.58 | Improved |
| add normal-map geometric supervision (ablation) | Masked LRM · qualitative mesh quality | surface quality | Improved |
| use normal loss instead of depth loss (ablation) | Masked LRM · qualitative mesh quality | surface quality | Improved |
| use 3D-consistent occlusion masking instead of 25% uniformly random patch masking (ablation) | Masked LRM · qualitative editing | edit sharpness | Improved |
| Masked LRM instead of Instant3Dit for text-guided editing | Instant3Dit · Mesh editing (ViT-L-14) | CLIP cosine similarity: 0.303 -> 0.323 | Improved |
| Masked LRM instead of MagicClay for text-guided editing | MagicClay · Mesh editing (ViT-BigG-14) | CLIP cosine similarity: 0.286 -> 0.337 | Improved |
| feed-forward Masked LRM instead of optimization-based TextDeformer | TextDeformer · runtime | runtime: ~20mins -> < 3sec | Improved |
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.
- Instant3Dit — Instant3dit: Multiview Inpainting for Fast Editing of 3D Objects (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
- PrEditor3D: Fast and Precise 3D Shape Editing (CVPR 2025 · similarity 0.90)
- EditVerse3D: High-Quality 3D Object Editing with Region-Aware Learning (ECCV 2026 · similarity 0.88)
- LRM-Zero: Training Large Reconstruction Models with Synthesized Data (NeurIPS 2024 · similarity 0.88)
- Instant3dit: Multiview Inpainting for Fast Editing of 3D Objects (CVPR 2025 · similarity 0.88)
- GTR: Improving Large 3D Reconstruction Models through Geometry and Texture Refinement (ICLR 2025 · similarity 0.87)
- Thinking in Boxes: 3D Editing in Real Images Made Easy (NeurIPS 2026 · 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.
