Research map / 3D vision / 3D generation
VividDreamer: Invariant Score Distillation for Hyper-Realistic Text-to-3D Generation
Wenjie Zhuo, Fan Ma, Hehe Fan, Yi Yang
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 |
|---|---|---|---|
| replace SDS reconstruction term with invariant score term | SDS · Text-to-3D, 50 prompts | average saturation: 203.75 -> 144.92 | Improved |
| increase classifier-free guidance scale in SDS (ablation) | SDS · 2D text-to-image analysis | over-saturation (qualitative) | Hurt |
| use medium guidance scale 7.5 with ISD instead of large scale 100 with SDS | ISD · text-to-3D | over-saturation (qualitative) | Improved |
| increase interval c from 0 to 20 (ablation) | ISD · Text-to-3D, 50 prompts | CLIP Score: 73.37 -> 81.06 | Improved |
| increase interval c from 20 to 50 (ablation) | ISD · Text-to-3D, 50 prompts | CLIP Score: 81.06 -> 80.87 | No clear effect |
| use ISD instead of ProlificDreamer | text-to-3D distillation · Text-to-3D, 50 prompts | CLIP Score: 78.90 -> 81.06 | Improved |
| use ISD instead of NFSD | text-to-3D distillation · User study (47 participants) | Image quality preference (%): 10.82 -> 61.84 | Improved |
| use lambda(t) weighting from derivation instead of constant 1 (ablation) | ISD · qualitative sushi car | text consistency (qualitative) | Improved |
| apply ISD to address multi-face problem | ISD · text-to-3D | Janus problem | No clear effect |
Closest papers by meaning
- Consistent3D: Towards Consistent High-Fidelity Text-to-3D Generation with Deterministic Sampling Prior (CVPR 2024 · similarity 0.90)
- Taming Mode Collapse in Score Distillation 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)
- Text-to-3D with Classifier Score Distillation (ICLR 2024 · similarity 0.89)
- Consistent Flow Distillation for Text-to-3D Generation (ICLR 2025 · similarity 0.89)
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.
