Research map / 3D vision / 3D generation
DreamCS: Geometry-Aware Text-to-3D Generation with Unpaired 3D Reward Supervision
Xiandong Zou, Ruihao Xia, Hongsong Wang, Pan Zhou
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 3D RewardCS guidance to DreamFusion | DreamFusion · GPTEval3D 110 prompts | GA: 2.53 -> 2.96 | Improved |
| add 3D RewardCS guidance to MVDream | MVDream · GPTEval3D 110 prompts | VisionReward: -3.31 -> -2.11 | Improved |
| add RewardCS guidance to Fantasia3D | Fantasia3D · GPTEval3D 110 prompts | VisionReward: -3.03 -> -1.01 | Improved |
| add 3D RewardCS in place of 2D Reward3D guidance | MVDream · MiniCPM-o evaluation, 3D plausibility | 3D plausibility: 3.35 -> 4.05 | Improved |
| add RewardCS in place of Reward3D guidance | MVDream · User study (60 prompts) | 3D plausibility: 3.15 -> 3.72 | Improved |
| add RewardCS guidance | MVDream · Janus artifact proportion | proportion with Janus artifacts: 0.52 -> 0.30 | Improved |
| combine 2D Reward3D and 3D RewardCS guidance (ablation) | DreamFusion · GPTEval3D 110 prompts | GA: 2.96 -> 3.22 | Improved |
| add CS-divergence separation loss (lambda=1 vs 0) (ablation) | RewardCS · MVDream GPTEval3D | GA: 1.97 -> 2.95 | Improved |
| raise CS-divergence weight from 1 to 10 (ablation) | RewardCS · MVDream GPTEval3D | GA: 2.95 -> 1.26 | Hurt |
| use progressive reward guidance weight instead of fixed low or high alpha (ablation) | DreamCS · MVDream | qualitative geometry | Improved |
| add Adaptive Mesh Fusion step (ablation) | DreamCS · geometric deviation | Chamfer distance | No clear effect |
Closest papers by meaning
- DreamAlign: Dynamic Text-to-3D Optimization with Human Preference Alignment (AAAI 2025 · similarity 0.93)
- DreamDPO: Aligning Text-to-3D Generation with Human Preferences via Direct Preference Optimization (ICML 2025 · similarity 0.92)
- Let Language Constrain Geometry: Vision–Language Models as Semantic and Spatial Critics for 3D Generation (ICML 2026 · similarity 0.88)
- MVReward: Better Aligning and Evaluating Multi-View Diffusion Models with Human Preferences (AAAI 2025 · similarity 0.87)
- Uni3D: Exploring Unified 3D Representation at Scale (ICLR 2024 · similarity 0.87)
- VP3D: Unleashing 2D Visual Prompt for Text-to-3D Generation (CVPR 2024 · 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.
