AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Easier Painting Than Thinking: Can Text-to-Image Models Set the Stage, but Not Direct the Play?

Ouxiang Li, Yuan Wang, Xinting Hu, Huijuan Huang, Rui Chen, Jiarong Ou, Xin Tao, Pengfei Wan, Xiaojuan Qi, Fuli Feng

ICLR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 31 citations (OpenAlex)

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 changedWhereMeasuredEffect
add intermediate reasoning and prompt rewriting (Think mode)BAGEL · T2I-CoReBench reasoning meanreasoning mean: 34.1 -> 41.9Improved
add intermediate reasoning and prompt rewriting (Think mode)BAGEL · T2I-CoReBench composition meancomposition mean: 46.4 -> 39.6Hurt
add o3 prompt rewriting (ablation)FLUX.1-Krea-dev · T2I-CoReBench reasoningreasoning mean: 44.3 -> 65.5Improved
add o3 prompt rewriting (ablation)Qwen-Image · T2I-CoReBench reasoningreasoning mean: 49.3 -> 72.7Improved
add o3 prompt rewriting (ablation)Nano Banana · T2I-CoReBench reasoningreasoning mean: 73.6 -> 78.2Improved
add o3 prompt rewriting (ablation)GPT-Image · T2I-CoReBench reasoningreasoning mean: 69.0 -> 79.9Improved
add o3 prompt rewriting (ablation)Nano Banana · T2I-CoReBench GR and ARGR / AR score: 83.2 (down 0.8), 81.4 (down 1.7)Hurt
add o3 prompt rewriting (ablation)GPT-Image · T2I-CoReBench PRPR score: 87.1 (down 0.2)No clear effect
add o3 prompt rewriting (ablation)FLUX.1-Krea-dev · T2I-CoReBench LRLR score: 30.3 -> 64.9Improved
switch from diffusion to autoregressive/unified paradigmopen-source T2I models · T2I-CoReBenchoverallMixed
gap between reasoning and composition in best closed modelNano Banana Pro · T2I-CoReBenchmean score: 89.7 -> 82.7Hurt
gap between reasoning and composition in best open modelQwen-Image-2512 · T2I-CoReBenchmean score: 83.7 -> 51.7Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy