Research map / 3D vision / 3D generation
DreamView: Injecting View-specific Text Guidance into Text-to-3D Generation
Junkai Yan, Yipeng Gao, Qize Yang, Xihan Wei, Xuansong Xie, Ancong Wu, WEI-SHI ZHENG
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 DreamView-2D adaptive guidance with view-specific and overall text instead of MVDream | DreamView-2D · Rendered validation set, view text | CLIP score: 28.6 -> 32.1 (vs MVDream with overall text) | Improved |
| DreamView-2D versus MVDream on overall text | DreamView-2D · Rendered validation set, overall text | CLIP score: 31.3 vs 31.1 | No clear effect |
| train with 3D data and view-specific text instead of 2D-only SD-v2.1 | DreamView-2D · Rendered validation set | Inception score: 15.3 -> 14.5 | Hurt |
| increase guidance margin (ablation) | DreamView-2D · Rendered validation set | CLIP score (overall vs view text) | Mixed |
| choose margin -0.025 or 0.0 rather than large margins (ablation) | DreamView-2D · qualitative, Iron Man example | customization vs consistency | Improved |
| replace Stable Diffusion with DreamView-2D in DreamFusion | DreamView-3D · text-to-3D user study (35 participants) | users choosing method: 74.5% chose DreamView-3D for text fit | Improved |
Closest papers by meaning
- ViewDiff: 3D-Consistent Image Generation with Text-to-Image Models (CVPR 2024 · similarity 0.89)
- DreamControl: Control-Based Text-to-3D Generation with 3D Self-Prior (CVPR 2024 · similarity 0.88)
- DreamFlow: High-quality text-to-3D generation by Approximating Probability Flow (ICLR 2024 · similarity 0.88)
- DreamDPO: Aligning Text-to-3D Generation with Human Preferences via Direct Preference Optimization (ICML 2025 · similarity 0.87)
- LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching (CVPR 2024 · similarity 0.87)
- ArtiScene: Language-Driven Artistic 3D Scene Generation Through Image Intermediary (CVPR 2025 · 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.
