Research map / Image generation / Text-to-image diffusion
TV-3DG: Mastering Text-to-3D Customized Generation with Visual Prompt
Jiahui Yang, Donglin Di, Baorui Ma, Jianxun Cui, Xun Yang, Yongjia Ma, Wenzhang Sun, Wei Chen, Zhou Xue, Meng Wang, Yebin Liu
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 SGC module (ablation) | TV-3DG · Text-to-3D | CLIP-Score ViT-B/16: 0.7879 -> 0.8000 | Improved |
| add SGC module (ablation) | TV-3DG · Stylize | CLIP-Score ViT-B/16: 0.7615 -> 0.7790 | Improved |
| add SGC module (ablation) | TV-3DG · Text-to-3D | FID: 364.2494 -> 351.5113 | Improved |
| add VPCSM visual prompt guidance on top of CSM (ablation) | TV-3DG · Text-to-3D | CLIP-Score ViT-B/16: 0.7514 -> 0.7879 | Improved |
| add VPCSM visual prompt guidance on top of CSM (ablation) | TV-3DG · Stylize | CLIP-Score ViT-B/16: 0.5611 -> 0.7615 | Improved |
| use CSM instead of SDS | TV-3DG framework · Text-to-3D | CLIP-Score ViT-B/16: 0.7675 -> 0.8000 | Improved |
| use CSM instead of SDS | TV-3DG framework · Text-to-3D | FID: 405.799 -> 351.511 | Improved |
| set tau to 0.5 instead of 1.0 (ablation) | TV-3DG · Text-to-3D | CLIP-Score ViT-B/16: 0.7473 -> 0.8000 | Improved |
| set lambda1 to 1 instead of 0 (ablation) | TV-3DG · Text-to-3D | CLIP-Score ViT-B/16: 0.7282 -> 0.8000 | Improved |
| use lambda2=5 instead of 4 (ablation) | TV-3DG · Text-to-3D | CLIP-Score ViT-L/14: 0.6587 -> 0.6685 | Mixed |
| set s=0 instead of 1 (ablation) | TV-3DG · Stylize | CLIP-Score ViT-B/16: 0.7790 -> 0.8023 | Mixed |
| use TV-3DG instead of VP3D | VP3D · Text-to-3D | CLIP-Score ViT-B/16: 0.7855 -> 0.8000 | 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.
- VP3D — VP3D: Unleashing 2D Visual Prompt for Text-to-3D Generation (CVPR 2024) · better on 2 of 2 shared results
- DreamFusion — DreamFusion: Text-to-3D using 2D Diffusion (ICLR 2023) · better on 1 of 1 shared result
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
- Text-to-3D with Classifier Score Distillation (ICLR 2024 · similarity 0.90)
- VP3D: Unleashing 2D Visual Prompt for Text-to-3D Generation (CVPR 2024 · similarity 0.90)
- LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching (CVPR 2024 · similarity 0.90)
- Rethinking Score Distilling Sampling for 3D Editing and Generation (ICML 2025 · similarity 0.89)
- Taming Mode Collapse in Score Distillation for Text-to-3D Generation (CVPR 2024 · similarity 0.88)
- DreamFlow: High-quality text-to-3D generation by Approximating Probability Flow (ICLR 2024 · 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.
