Research map / 3D vision / 3D generation
MiDSummer: Multi-Guidance Diffusion for Controllable Zero-Shot Immersive Gaussian Splatting Scene Generation
Anjun Hu, Richard Tomsett, Valentin Gourmet, Massimo Camplani, Jas Kandola, Hanting Xie
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 GDM guidance to LLM-only layout | LLM-only layout · SUN RGB-D text-to-layout | L/R accuracy: 47.5 -> 88.8 | Improved |
| add LLM guidance to GDM-only layout | GDM-only layout · SUN RGB-D text-to-layout | Volume error: 255.5 -> 50.8 | Improved |
| add LLM guidance to GDM-only layout | GDM-only layout · SUN RGB-D text-to-layout | Bi/Sm accuracy: 76.4 -> 88.8 | Improved |
| add LLM guidance to GDM-only layout | GDM-only layout · SUN RGB-D text-to-layout | U/D accuracy: 89.0 -> 87.4 | Hurt |
| add LLM guidance to GDM-only layout (scene graph input) | GDM-only layout · SUN RGB-D scene-graph-to-layout | Volume error: 332.7 -> 64.2 | Improved |
| add LLM guidance to GDM-only layout (scene graph input) | GDM-only layout · SUN RGB-D scene-graph-to-layout | L/R accuracy: 89.5 -> 84.0 | Hurt |
| add dual-guidance layout planning (ablation) | MiDSummer · text-to-scene generation | KID vs SUN RGB-D: 0.1198 -> 0.0647 | Improved |
| add dual-guidance layout planning (ablation) | MiDSummer · text-to-scene generation | VarLap: 615.4 -> 633.8 | Improved |
| add layout-aware refinement (ablation) | MiDSummer · text-to-scene generation | VarLap: 545.9 -> 633.8 | Improved |
| add layout-aware refinement (ablation) | MiDSummer · text-to-scene generation | KID vs SUN RGB-D: 0.0569 -> 0.0647 | Hurt |
| use equal 0.5/0.5 GDM/LLM weights instead of GDM only (ablation) | layout diffusion · SUN RGB-D text-to-layout | Mean binary relationship accuracy: 84.22 -> 71.61 | Hurt |
| shift weight from GDM to LLM (1.0/0.0 to 0.0/1.0) (ablation) | layout diffusion · SUN RGB-D text-to-layout | Volume error: 255.5 -> 64.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GALA3D — GALA3D: Towards Text-to-3D Complex Scene Generation via Layout-guided Generative Gaussian Splatting (ICML 2024) · better on 3 of 3 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
- EditSplat: Multi-View Fusion and Attention-Guided Optimization for View-Consistent 3D Scene Editing with 3D Gaussian Splatting (CVPR 2025 · similarity 0.89)
- CoT-lized Diffusion: Let's Reinforce T2I Generation Step-by-step (NeurIPS 2025 · similarity 0.88)
- SplatFlow: Multi-View Rectified Flow Model for 3D Gaussian Splatting Synthesis (CVPR 2025 · similarity 0.88)
- X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllability (NeurIPS 2025 · similarity 0.87)
- Taming Video Diffusion Prior with Scene-Grounding Guidance for 3D Gaussian Splatting from Sparse Inputs (CVPR 2025 · similarity 0.87)
- BloomScene: Lightweight Structured 3D Gaussian Splatting for Crossmodal Scene Generation (AAAI 2025 · 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.
