Research map / Image generation / Text-to-image diffusion
Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models
Zengbin Wang, Xuecai Hu, Yong Wang, Feng Xiong, Man Zhang, Xiangxiang Chu
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 |
|---|---|---|---|
| fine-tune on SpatialT2I | SD-XL · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy: 41.2 -> 45.4 | Improved |
| fine-tune on SpatialT2I | UniWorld-V1 · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy: 54.2 -> 59.9 | Improved |
| fine-tune on SpatialT2I | OmniGen2 · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy: 56.4 -> 60.8 | Improved |
| fine-tune on SpatialT2I | UniWorld-V1 · SpatialGenEval (Qwen2.5-VL-72B judge) (Comparison) | accuracy: 26.1 -> 34.6 | Improved |
| fine-tune on SpatialT2I | UniWorld-V1 · SpatialGenEval (Qwen2.5-VL-72B judge) (Motion) | accuracy: 69.6 -> 66.0 | Hurt |
| fine-tune on SpatialT2I | OmniGen2 · SpatialGenEval (Qwen2.5-VL-72B judge) (Object) | accuracy: 51.5 -> 61.8 | Improved |
| use higher-scoring source model subsets for SFT data (Unipic-v2 54.3, Bagel 57.0, Qwen-Image 60.6) (ablation) | SD-XL and OmniGen2 · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy | Improved |
| scale SFT data from 0% to 100% by adding better subsets (ablation) | OmniGen2 · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy | Improved |
| use T5 text encoder (SD-3) instead of CLIP only (SD-1.5) | SD-3 / SD-1.5 · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy: 28.5 -> 54.6 | Improved |
| scale diffusion model from 8B SD-3.5-L to 20B Qwen-Image | Qwen-Image · SpatialGenEval (Qwen2.5-VL-72B judge) | overall accuracy: 54.0 -> 60.6 | Improved |
| use GPT-4o instead of Qwen2.5-VL-72B as judge | SpatialGenEval · SpatialGenEval top-5 models | model ranking: 60.8 vs 60.6 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- NextStep-1 — NextStep-1: Toward Autoregressive Image Generation with Continuous Tokens at Scale (ICLR 2026) · better on 1 of 1 shared result
- Infinity — Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesis (CVPR 2025) · better on 1 of 1 shared result
- PixArt-sigma — PixArt-Sigma: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generation (ECCV 2024) · better on 1 of 1 shared result
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
- Exploring Spatial Intelligence from a Generative Perspective (CVPR 2026 · similarity 0.92)
- MMSI-Bench: A Benchmark for Multi-Image Spatial Intelligence (ICLR 2026 · similarity 0.92)
- Is A Picture Worth A Thousand Words? Delving Into Spatial Reasoning for Vision Language Models (NeurIPS 2024 · similarity 0.91)
- SpatialScore: Towards Comprehensive Evaluation for Spatial Intelligence (CVPR 2026 · similarity 0.91)
- SpatiaLab: Can Vision–Language Models Perform Spatial Reasoning in the Wild? (ICLR 2026 · similarity 0.90)
- Stitch and Tell: A Structured Data Augmentation Method for Spatial Understanding (NeurIPS 2025 · similarity 0.90)
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.
