Research map / Image generation / Text-to-image diffusion
DreamFusion: Text-to-3D using 2D Diffusion
Ben Poole, Ajay Jain, Jonathan T. Barron, Ben Mildenhall
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 Score Distillation Sampling instead of minimizing the diffusion training loss | Differentiable image parameterizations · 2D image sampling | sample realism | Improved |
| Add large ranges of viewpoints (ViewAug) (ablation) | DreamFusion · Object-centric COCO | geometry quality | Improved |
| Add view-dependent prompts (ViewDep) (ablation) | DreamFusion · Object-centric COCO | geometry quality | Improved |
| Add lighting to illuminated renders (ablation) | DreamFusion · Object-centric COCO | geometry quality | Improved |
| Optimize textureless shaded geometry images (ablation) | DreamFusion · Object-centric COCO | geometry quality | Improved |
| Add the full set of optimization choices: ViewAug, ViewDep, Lighting, and Textureless (ablation) | DreamFusion · Object-centric COCO | CLIP R-Precision on full renderings: +12.5% | Improved |
| Evaluate only albedo renders instead of checking geometry with textureless renders (ablation) | DreamFusion ablation evaluation · Object-centric COCO | geometry quality | Hurt |
| Use a large classifier-free guidance weight | DreamFusion · Text-to-3D synthesis | sample quality | Improved |
| Use a uniform timestep weighting instead of w(t)=σ_t² | Score Distillation Sampling · Text-to-3D synthesis | sample quality | No clear effect |
Later papers that compared with this method
15 later accepted papers report DreamFusion as a baseline in their results tables; 15 of them report a better number on most of the results they share with it.
- SegmentDreamer: Towards High-fidelity Text-to-3D Synthesis with Segmented Consistency Trajectory Distillation (ICCV 2025 · better on 7 of 7 shared results)
- LEGO-Maker: A Semantic-Driven Algorithm for Text-to-3D Generation (ICCV 2025 · better on 4 of 4 shared results)
- GPS as a Control Signal for Image Generation (CVPR 2025 · better on 3 of 3 shared results)
- Rethinking Score Distilling Sampling for 3D Editing and Generation (ICML 2025 · better on 2 of 2 shared results)
- TV-3DG: Mastering Text-to-3D Customized Generation with Visual Prompt (TPAMI 2025 · better on 1 of 1 shared result)
- Text-to-3D with Classifier Score Distillation (ICLR 2024 · better on 4 of 5 shared results)
- CAD: Photorealistic 3D Generation via Adversarial Distillation (CVPR 2024 · better on 4 of 4 shared results)
- DiSR-NeRF: Diffusion-Guided View-Consistent Super-Resolution NeRF (CVPR 2024 · better on 4 of 4 shared results)
7 more not listed.
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
- Let 2D Diffusion Model Know 3D-Consistency for Robust Text-to-3D Generation (ICLR 2024 · similarity 0.93)
- DreamTime: An Improved Optimization Strategy for Diffusion-Guided 3D Generation (ICLR 2024 · similarity 0.92)
- SwiftBrush: One-Step Text-to-Image Diffusion Model with Variational Score Distillation (CVPR 2024 · similarity 0.92)
- Instant3D: Fast Text-to-3D with Sparse-view Generation and Large Reconstruction Model (ICLR 2024 · similarity 0.91)
- IT3D: Improved Text-to-3D Generation with Explicit View Synthesis (AAAI 2024 · similarity 0.91)
- DreamFlow: High-quality text-to-3D generation by Approximating Probability Flow (ICLR 2024 · similarity 0.91)
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.
