Research map / 3D vision / 3D generation
Incremental Scene Synthesis
Benjamin Planche, Xuejian Rong, Ziyan Wu, Srikrishna Karanam, Harald Kosch, YingLi Tian, Jan Ernst, ANDREAS HUTTER
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 incremental global memory and scene hallucination instead of GTM-SM | CelebA and HoME-2D agents · 2D scene synthesis | Hallucination SSIM: 0.41 -> 0.72 | Improved |
| Use incremental global memory and scene hallucination instead of GTM-SM | AVD · AVD | Anamnesis L1: 0.37 -> 0.22 | Improved |
| Use observation-based localization rather than GTM-SM localization | Doom · Doom | APE mean: 2.15u -> 1.64u | Improved |
| Use observation-based localization rather than GTM-SM localization | Doom · Doom | ATE mean: 1.81u -> 1.95u | Hurt |
| Use observation-based localization rather than GTM-SM localization | AVD · AVD | APE mean: 0.77u -> 0.32u | Improved |
| Remove localization from the full pipeline (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.08 -> 0.14 | Hurt |
| Remove culling from the full pipeline (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.08 -> 0.08 | No clear effect |
| Remove the hallucinatory DAE from the full pipeline (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Hallucination SSIM: 0.70 -> 0.66 | Hurt |
| Remove the encoder from the full pipeline (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.08 -> 0.13 | Improved |
| Remove encoder and localization (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.08 -> 0.14 | Hurt |
| Remove encoder, localization, DAE, and culling (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.08 -> 0.18 | Hurt |
| Add culling to encoder and localization (ablation) | Incremental Scene Synthesis · CelebA A^c_cel | Anamnesis L1: 0.15 -> 0.08 | Improved |
Closest papers by meaning
- SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent (NeurIPS 2025 · similarity 0.87)
- SimVS: Simulating World Inconsistencies for Robust View Synthesis (CVPR 2025 · similarity 0.86)
- Habitat Synthetic Scenes Dataset (HSSD-200): An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation (CVPR 2024 · similarity 0.86)
- Hallo3D: Multi-Modal Hallucination Detection and Mitigation for Consistent 3D Content Generation (NeurIPS 2024 · similarity 0.85)
- X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllability (NeurIPS 2025 · similarity 0.85)
- Video Perception Models for 3D Scene Synthesis (NeurIPS 2025 · similarity 0.85)
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.
