AI Paper Atlas

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

T2I-CompBench++: An Enhanced and Comprehensive Benchmark for Compositional Text-to-Image Generation

Kaiyi Huang, Chengqi Duan, Kaiyue Sun, Enze Xie, Zhenguo Li, Xihui Liu

TPAMI 2025 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 209 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
fine-tune with reward-selected samples (GORS)Stable v2 · ColorB-VQA: 0.5065 -> 0.6603Improved
fine-tune with reward-selected samples (GORS)Stable v2 · ShapeB-VQA: 0.4221 -> 0.4785Improved
fine-tune with reward-selected samples (GORS)Stable v2 · 2D-SpatialUniDet: 0.1342 -> 0.1815Improved
fine-tune with reward-selected samples (GORS)Stable v2 · Complex3-in-1: 0.3386 -> 0.3328No clear effect
add Attend-and-ExciteStable v2 · ColorB-VQA: 0.5065 -> 0.6400Improved
add Composable DiffusionStable v2 · ColorB-VQA: 0.5065 -> 0.4063Hurt
add Structured DiffusionStable v2 · ColorB-VQA: 0.5065 -> 0.4990No clear effect
also fine-tune CLIP text encoder (not U-Net only) (ablation)GORS · ColorB-VQA: 0.6216 -> 0.6570Improved
also fine-tune U-Net (not CLIP only) (ablation)GORS · ColorB-VQA: 0.5507 -> 0.6570Improved
use full reward threshold instead of half (ablation)GORS · ColorB-VQA: 0.6157 -> 0.6570Improved
use sample selection threshold instead of zero threshold (ablation)GORS · ColorB-VQA: 0.6130 -> 0.6570Improved
scale training prompts from 25 to 1400 (ablation)GORS · Complex3-in-1: 0.2596 -> 0.3504Improved

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