T2I-CompBench

'T-2I-CompBench' is a benchmark used to evaluate text-image alignment in multimodal diffusion transformers by assessing aspects such as object appearance, attribute binding, and spatial relationships.

Papers using T2I-CompBench (8)

IV-CoT: Implicit Visual Chain-of-Thought for Structure-Aware Text-to-Image Generation2026Can AI Draw Science? A Benchmark for Evaluating Scientific Figure Generation by Text-to-Image and Multimodal Models2026Prompt Reinjection: Alleviating Prompt Forgetting in Multimodal Diffusion Transformers for Text-to-Image Generation2026Iterative Refinement Improves Compositional Image Generation2026Shape of Thought: Progressive Object Assembly via Visual Chain-of-Thought2026SRUM: Fine-Grained Self-Rewarding for Unified Multimodal Models2025Rethinking Cross-Modal Interaction in Multimodal Diffusion Transformers2025REVISION: Rendering Tools Enable Spatial Fidelity in Vision-Language Models2024 · 2 cites