Research map / 3D vision / 3D generation
DreamPartGen: Semantically Grounded Part-Level 3D Generation via Collaborative Latent Denoising
Tianjiao (Joey) Yu, Xinzhuo Li, Muntasir Wahed, Jerry Xiong, Yifan Shen, Ying Shen, Ismini Lourentzou
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 |
|---|---|---|---|
| remove global relational tokens S_glb and object-level sync (add them) (ablation) | DreamPartGen · PartRel3D subset | CD: 0.292 -> 0.145 | Improved |
| add local semantic tokens S_loc co-denoised with part latents (ablation) | DreamPartGen · PartRel3D subset | CD: 0.781 -> 0.145 | Improved |
| add learnable part identifier embedding (ablation) | DreamPartGen · PartRel3D subset | IoU: 0.438 -> 0.212 | Improved |
| add persistent RSLs for text-shape alignment (ablation) | DreamPartGen · PartRel3D subset | ULIP-T: 0.084 -> 0.158 | Improved |
| add spatial triplets to text conditioning (ablation) | DreamPartGen · PartRel3D | CD: 0.167 -> 0.112 | Improved |
| add functional triplets alone instead of text (ablation) | DreamPartGen · PartRel3D | CD: 0.167 -> 0.180 | Hurt |
| add image to Text+FT+ST (ablation) | DreamPartGen · PartRel3D | CD: 0.085 -> 0.080 | Improved |
| increase local RSL tokens beyond 16 (ablation) | DreamPartGen · PartRel3D | CD: 0.084 -> 0.085 | No clear effect |
| use VLM-parsed relations at inference instead of prompt-only (ablation) | DreamPartGen · PartRel3D | CD: 0.099 vs 0.101 (Qwen2.5-VL) | No clear effect |
| test on held-out relation predicates | DreamPartGen · PartRel3D OOD-rel | r-FID: 9.783 -> 10.631 | Mixed |
| DreamPartGen vs PartCrafter on ShapeNet part independence | DreamPartGen · ShapeNet | IoU: 0.499 vs 0.503 | No clear effect |
| part-aware co-denoising vs baselines on text-shape alignment | DreamPartGen · PartVerse object-level | ULIP-T: 0.164 -> 0.197 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PartCrafter — PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025) · better on 6 of 7 shared results
- HoloPart — HoloPart: Generative 3D Part Amodal Segmentation (ICLR 2026) · better on 5 of 5 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
- PartGen: Part-level 3D Generation and Reconstruction with Multi-view Diffusion Models (CVPR 2025 · similarity 0.91)
- PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformers (NeurIPS 2025 · similarity 0.90)
- AutoPartGen: Autoregressive 3D Part Generation and Discovery (NeurIPS 2025 · similarity 0.89)
- Let Language Constrain Geometry: Vision–Language Models as Semantic and Spatial Critics for 3D Generation (ICML 2026 · similarity 0.89)
- PartNeXt: A Next-Generation Dataset for Fine-Grained and Hierarchical 3D Part Understanding (NeurIPS 2025 · similarity 0.88)
- Part-X-MLLM: Part-aware 3D Multimodal Large Language Model (ICLR 2026 · similarity 0.88)
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.
