Research map / 3D vision / 3D generation
Think-Then-Generate: Structural Chain-of-Thought Reasoning for Consistent 3D Generation
Xinyue Liu, Jin Liu, Hongbo Wang, Ran He, Huaibo Huang
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 3DBlueprint-CoT (staged prompt planning) (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/32: 22.76 -> 24.65 | Improved |
| add 3DBlueprint-CoT (staged prompt planning) (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/16: 26.86 -> 29.15 | Improved |
| add 3DRefine-CoT (reflection, correction, MLLM voting) (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/32: 23.56 -> 24.65 | Improved |
| add 3DRefine-CoT (reflection, correction, MLLM voting) (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/16: 28.87 -> 29.15 | Improved |
| add Cross-view Semantic Appearance Alignment (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/32: 23.97 -> 24.65 | Improved |
| add Cross-view Semantic Appearance Alignment (ablation) | Thoughtful3D on GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score L/14: 29.42 -> 30.39 | Improved |
| add 3DBlueprint-CoT alone relative to baseline (w/o A vs Baseline) (ablation) | GaussianDreamer · Text-to-3D (100 prompts) | CLIP Score B/16: 26.71 -> 26.86 | No clear effect |
| add Thoughtful3D to Fantasia3D | Fantasia3D · Text-to-3D (100 prompts) | CLIP Score B/32: 16.71 -> 23.33 | Improved |
| add Thoughtful3D to Magic3D | Magic3D · Text-to-3D (100 prompts) | CLIP Score B/32: 16.28 -> 19.64 | Improved |
| add Thoughtful3D to DreamFusion-IF | DreamFusion-IF · Text-to-3D (100 prompts) | CLIP Score L/14: 23.22 -> 28.67 | Improved |
| add Thoughtful3D to Zero123 | Zero123 · Image-to-3D (50 objects) | Volume IoU: 0.3203 -> 0.3714 | Improved |
| add Thoughtful3D to Wonder3D | Wonder3D · Image-to-3D (50 objects) | SSIM: 0.8201 -> 0.8234 | 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.
- GaussianDreamer — GaussianDreamer: Fast Generation from Text to 3D Gaussians by Bridging 2D and 3D Diffusion Models (CVPR 2024) · better on 3 of 3 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
- Ouroboros3D: Image-to-3D Generation via 3D-aware Recursive Diffusion (CVPR 2025 · similarity 0.88)
- Let Language Constrain Geometry: Vision–Language Models as Semantic and Spatial Critics for 3D Generation (ICML 2026 · similarity 0.88)
- Eval3D: Interpretable and Fine-grained Evaluation for 3D Generation (CVPR 2025 · similarity 0.88)
- Shape of Thought: Progressive Object Assembly via Visual Chain-of-Thought (ICML 2026 · similarity 0.87)
- Sculpt3D: Multi-View Consistent Text-to-3D Generation with Sparse 3D Prior (CVPR 2024 · similarity 0.87)
- Think with 3D: Geometric Imagination Grounded Spatial Reasoning from Limited Views (CVPR 2026 · similarity 0.87)
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.
