Research map / 3D vision / 3D generation
Score Jacobian Chaining: Lifting Pretrained 2D Diffusion Models for 3D Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 SJC instead of Stable-DreamFusion | Stable Diffusion-guided text-prompted 3D generation · Text-prompted 3D generation | 3D generation quality | Helped |
| Add Perturb-and-Average Scoring to estimate scores on rendered images | SJC · Rendered images used for 3D optimization | score guidance quality | Helped |
| Use an annealed σ schedule instead of a random σ schedule (ablation) | PAAS image canvas sampling with FFHQ DDPM · FFHQ | sample quality and pose variation | Helped |
| Use an annealed σ schedule instead of a random σ schedule (ablation) | PAAS image canvas sampling with LSUN Bedroom DDPM · LSUN Bedroom | sample quality | Helped |
| Use a random σ schedule instead of an annealed σ schedule (ablation) | PAAS image canvas sampling with Stable Diffusion at guidance scale 10.0 · Text-conditioned image canvas sampling | sample quality | Helped |
| Use a random σ schedule instead of an annealed σ schedule (ablation) | PAAS image canvas sampling with Stable Diffusion at guidance scale 3.0 · Text-conditioned image canvas sampling | sample quality | Hurt |
| Add an emptiness loss (ablation) | SJC 3D generation · Text-prompted 3D generation | geometry quality | Helped |
| Schedule the emptiness-loss weight rather than keeping it constant (ablation) | SJC with emptiness loss · Text-prompted 3D generation | geometry quality | Helped |
| Add a moderate center depth loss (ablation) | SJC 3D generation · Text-prompted 3D generation | object placement and scale | Helped |
Closest papers by meaning
- Diffusion$^2$: Dynamic 3D Content Generation via Score Composition of Video and Multi-view Diffusion Models (ICLR 2025 · similarity 0.88)
- A 3D Scene is Worth 1K Tokens: 3D-Grounded Representation for Scene Generation at Scale (NeurIPS 2026 · similarity 0.88)
- CAD: Photorealistic 3D Generation via Adversarial Distillation (CVPR 2024 · similarity 0.88)
- DreamCraft3D: Hierarchical 3D Generation with Bootstrapped Diffusion Prior (ICLR 2024 · similarity 0.88)
- Denoising Diffusion via Image-Based Rendering (ICLR 2024 · similarity 0.87)
- DreamTime: An Improved Optimization Strategy for Diffusion-Guided 3D Generation (ICLR 2024 · 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.