Research map / 3D vision / Face and shape
MeshPad: Interactive Sketch-Conditioned Artist-Reminiscent Mesh Generation and Editing
Haoxuan Li, Ziya Erkoç, Lei Li, Daniele Sirigatti, Vladislav Rosov, Angela Dai, Matthias Nießner
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 vertex-aligned speculator (ablation) | MeshPad addition network · IKEA (speculator ablation) | tokens per second: 60.7 -> 131.1 | Improved |
| add vertex-aligned speculator (ablation) | MeshPad addition network · IKEA (speculator ablation) | Chamfer distance (x0.001): 7.66 -> 6.78 | Improved |
| align speculator with vertex tokens (predict only y,z) (ablation) | MeshPad speculator · IKEA (speculator ablation) | Chamfer distance (x0.001): 9.00 -> 6.78 | Improved |
| align speculator with vertex tokens (ablation) | MeshPad speculator · IKEA (speculator ablation) | CLIP similarity: 94.43 -> 96.67 | Improved |
| align speculator with vertex tokens (ablation) | MeshPad speculator · IKEA (speculator ablation) | tokens per second: 138.9 -> 131.1 | No clear effect |
| train speculator jointly with the transformer instead of frozen (ablation) | MeshPad speculator · IKEA (speculator ablation) | Chamfer distance (x0.001): 57.13 -> 6.78 | Improved |
| train speculator jointly with the transformer (ablation) | MeshPad speculator · IKEA (speculator ablation) | FID: 211.46 -> 29.67 | Improved |
| use MeshPad instead of LAS/SENS baselines | MeshPad · ShapeNet sketch-conditioned generation | Chamfer distance (x0.001): 8.95 -> 6.20 | Improved |
| post-process LAS with MeshAnythingV2 | LAS · ShapeNet sketch-conditioned generation | Chamfer distance (x0.001): 15.02 -> 22.06 | Hurt |
| post-process LAS with MeshAnythingV2 | LAS · ShapeNet sketch-conditioned generation | FID: 46.90 -> 18.52 | Improved |
| post-process SENS with MeshAnythingV2 | SENS · ShapeNet sketch-conditioned generation | Chamfer distance (x0.001): 8.95 -> 29.43 | Hurt |
| post-process SENS with MeshAnythingV2 | SENS · ShapeNet sketch-conditioned generation | FID: 81.88 -> 42.93 | Improved |
Closest papers by meaning
- MeshGPT: Generating Triangle Meshes with Decoder-Only Transformers (CVPR 2024 · similarity 0.88)
- MeshAnything: Artist-Created Mesh Generation with Autoregressive Transformers (ICLR 2025 · similarity 0.88)
- Diff3DS: Generating View-Consistent 3D Sketch via Differentiable Curve Rendering (ICLR 2025 · similarity 0.87)
- MeshArt: Generating Articulated Meshes with Structure-Guided Transformers (CVPR 2025 · similarity 0.87)
- TreeMeshGPT: Artistic Mesh Generation with Autoregressive Tree Sequencing (CVPR 2025 · similarity 0.86)
- Instant3dit: Multiview Inpainting for Fast Editing of 3D Objects (CVPR 2025 · 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.
