Research map / 3D vision / 3D generation
Map2World: Segment Map Conditioned Text to 3D World Generation
Jaeyoung Chung, Suyoung Lee, Jianfeng Xiang, Jiaolong Yang, Kyoung Mu Lee
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 concatenation-MLP conditioning instead of IP-Adapter (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | FID (DINOv2): 80.81 -> 32.67 | Improved |
| use concatenation-MLP conditioning instead of IP-Adapter (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | PSNR: 20.28 -> 22.53 | Improved |
| use concatenation-MLP conditioning instead of IP-Adapter (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | LPIPS: 0.2499 -> 0.2137 | Improved |
| avoid classifier-free guidance in fine-tuning and sampling (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | FID (CLIP): 21.32 -> 11.79 | Improved |
| avoid classifier-free guidance in fine-tuning and sampling (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | PSNR: 21.95 -> 22.53 | Improved |
| add SLAT decoder fine-tuning on small scene cubes (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | FID (CLIP): 13.17 -> 11.79 | Improved |
| add SLAT decoder fine-tuning on small scene cubes (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | PSNR: 22.08 -> 22.53 | Improved |
| add SLAT decoder fine-tuning on small scene cubes (ablation) | detail enhancer · Detail enhancer ablation (1,500 test cubes) | FID (Inception v3): 17.89 -> 16.98 | Improved |
| add spectral (3D FFT) parameterization of sparse structure for initial noise optimization (ablation) | scale-aware initial latent optimization · Scale optimization with learning rate 9.0 | IoU and Dice: approximately 0.9 on average within five optimization steps | Improved |
| lower learning rate to 1.0 without spectral parameterization (ablation) | initial latent optimization · Scale optimization | convergence speed: avoids divergence but requires significantly more optimization steps | Mixed |
| replace SynCity with Map2World | world generator · World generation (35 NuiScene43 captions) | GPTscore average: 7.48 -> 7.93 | Improved |
| replace SynCity with Map2World | world generator · World generation (35 NuiScene43 captions) | World Quality (WQ): 7.25 -> 7.76 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SynCity — SynCity: Training-Free Generation of 3D Worlds (ICCV 2025) · better on 5 of 6 shared results
- GaussianCube — GaussianCube: A Structured and Explicit Radiance Representation for 3D Generative Modeling (NeurIPS 2024) · better on 5 of 5 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
- Let Language Constrain Geometry: Vision–Language Models as Semantic and Spatial Critics for 3D Generation (ICML 2026 · similarity 0.88)
- WorldCoder-Bench: Benchmarking Physically Grounded 3D World Synthesis (NeurIPS 2026 · similarity 0.87)
- Layout-your-3D: Controllable and Precise 3D Generation with 2D Blueprint (ICLR 2025 · similarity 0.87)
- CoT-lized Diffusion: Let's Reinforce T2I Generation Step-by-step (NeurIPS 2025 · similarity 0.86)
- 3D Space as a Scratchpad for Editable Text-to-Image Generation (CVPR 2026 · similarity 0.86)
- SSMG: Spatial-Semantic Map Guided Diffusion Model for Free-Form Layout-to-Image Generation (AAAI 2024 · 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.
