Research map / 3D vision / 3D generation
Connecting Consistency Distillation to Score Distillation for Text-to-3D Generation
Zongrui Li, Minghui Hu, Qian Zheng, Xudong Jiang
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 Brightness-Equalized Generation (BEG) to 3DGS rendering during optimization (ablation) | GCS / LucidDreamer · qualitative ablation, text-to-3D | over-saturation (visual) | Improved |
| add BEG to LucidDreamer (ablation) | LucidDreamer · qualitative ablation, text-to-3D | fidelity (visual) | No clear effect |
| add pixel-domain constraint L_CP (ablation) | L_CC + L_CG · qualitative ablation, text-to-3D | color distortion / saturation (visual) | Mixed |
| use compact consistency loss with midpoint e>0 instead of e=0 (CDS-like error bound) (ablation) | L_CG + L_CC · qualitative ablation, text-to-3D | color distortion (visual) | Improved |
| add Conditional Guidance loss in place of interval score distillation (ablation) | LucidDreamer · qualitative ablation, text-to-3D | generation quality (visual) | Improved |
| increase time interval delta from 100 (ablation) | GCS · qualitative, strawberry example | detail (visual) | Hurt |
| use second-order DPM-Solver instead of first-order (ablation) | GCS + BEG · qualitative, low CFG | illumination and detail (visual) | No clear effect |
| apply CFG in only one denoising step rather than every step (ablation) | Compact Consistency loss · text-to-3D low CFG weight | generation quality (visual) | Improved |
| GCS + BEG vs LucidDreamer on FID to Stable Diffusion 2.1 images | GCS + BEG · DreamFusion gallery, 30 prompts | FID: 109.57 -> 103.40 | Improved |
| GCS + BEG vs ProlificDreamer on CLIP score | GCS + BEG · DreamFusion gallery, 30 prompts | CLIP Score: 33.17 -> 32.37 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- LucidDreamer — LucidDreamer: Towards High-Fidelity Text-to-3D Generation via Interval Score Matching (CVPR 2024) · better on 5 of 5 shared results
- GaussianDreamer — GaussianDreamer: Fast Generation from Text to 3D Gaussians by Bridging 2D and 3D Diffusion Models (CVPR 2024) · better on 2 of 2 shared results
- DreamGaussian — DreamGaussian: Generative Gaussian Splatting for Efficient 3D Content Creation (ICLR 2024) · better on 2 of 2 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
- Consistent Flow Distillation for Text-to-3D Generation (ICLR 2025 · similarity 0.91)
- Consistent3D: Towards Consistent High-Fidelity Text-to-3D Generation with Deterministic Sampling Prior (CVPR 2024 · similarity 0.91)
- 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 0.89)
- Rethinking Score Distilling Sampling for 3D Editing and Generation (ICML 2025 · similarity 0.89)
- Text-to-3D using Gaussian Splatting (CVPR 2024 · 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.
