Research map / 3D vision / 3D generation
WorldAgents: Can Foundation Image Models be Agents for 3D World Models?
Ziya Erkoç, Angela Dai, Matthias Niessner
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 2D and 3D verifier agent (ablation) | Generator only · Ablation | CLIP Score: 19.07 -> 20.24 | Improved |
| add 2D and 3D verifier agent (ablation) | Generator only · Ablation | Inception Score: 2.23 -> 2.43 | Improved |
| add VLM director agent (ablation) | Generator + Verifier · Ablation | CLIP Score: 20.24 -> 21.80 | Improved |
| add VLM director agent (ablation) | Generator + Verifier · Ablation | Inception Score: 2.43 -> 2.94 | Improved |
| add 3D-aware sequential inpainting (ablation) | Generator + Verifier + Director · Ablation | CLIP Score: 21.80 -> 26.79 | Improved |
| add 3D-aware sequential inpainting (ablation) | Generator + Verifier + Director · Ablation | CLIP-IQA: 0.69 -> 0.89 | Improved |
| add 3D-aware sequential inpainting (ablation) | Generator + Verifier + Director · Ablation | Inception Score: 2.94 -> 2.26 | Hurt |
| full WorldAgents vs WorldExplorer | WorldAgents · 3D world generation | CLIP Score: 24.49 -> 26.79 | Improved |
| full WorldAgents vs Text2Room | WorldAgents · 3D world generation | Inception Score: 2.79 -> 2.26 | Hurt |
| replace GPT-4.1 with Qwen3-VL 8B as VLM | WorldAgents (Flux.2 Pro) · 3D world generation | CLIP Score: 26.79 -> 24.49 | Hurt |
| replace Flux.2 Pro with Flux.2 Klein 9B (GPT-4.1) | WorldAgents · 3D world generation | CLIP-IQA: 0.89 -> 0.84 | Hurt |
Closest papers by meaning
- 3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code (NeurIPS 2026 · similarity 0.88)
- VisualAgentBench: Towards Large Multimodal Models as Visual Foundation Agents (ICLR 2025 · similarity 0.87)
- IR3D-Bench: Evaluating Vision-Language Model Scene Understanding as Agentic Inverse Rendering (NeurIPS 2025 · similarity 0.87)
- WorldCoder-Bench: Benchmarking Physically Grounded 3D World Synthesis (NeurIPS 2026 · similarity 0.87)
- 4DWorldBench: A Comprehensive Evaluation Framework for 3D/4D World Generation Models (CVPR 2026 · similarity 0.87)
- Muses: 3D-Controllable Image Generation via Multi-Modal Agent Collaboration (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.
