CoT-lized Diffusion: Let's Reinforce T2I Generation Step-by-step
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911041903919104 |
|---|---|
| author | Liu, Zheyuan Ning, Munan Zhang, Qihui Yang, Shuo Wang, Zhongrui Yang, Yiwei Xu, Xianzhe Song, Yibing Chen, Weihua Wang, Fan Yuan, Li |
| author_facet | Liu, Zheyuan Ning, Munan Zhang, Qihui Yang, Shuo Wang, Zhongrui Yang, Yiwei Xu, Xianzhe Song, Yibing Chen, Weihua Wang, Fan Yuan, Li |
| contents | Current text-to-image (T2I) generation models struggle to align spatial composition with the input text, especially in complex scenes. Even layout-based approaches yield suboptimal spatial control, as their generation process is decoupled from layout planning, making it difficult to refine the layout during synthesis. We present CoT-Diff, a framework that brings step-by-step CoT-style reasoning into T2I generation by tightly integrating Multimodal Large Language Model (MLLM)-driven 3D layout planning with the diffusion process. CoT-Diff enables layout-aware reasoning inline within a single diffusion round: at each denoising step, the MLLM evaluates intermediate predictions, dynamically updates the 3D scene layout, and continuously guides the generation process. The updated layout is converted into semantic conditions and depth maps, which are fused into the diffusion model via a condition-aware attention mechanism, enabling precise spatial control and semantic injection. Experiments on 3D Scene benchmarks show that CoT-Diff significantly improves spatial alignment and compositional fidelity, and outperforms the state-of-the-art method by 34.7% in complex scene spatial accuracy, thereby validating the effectiveness of this entangled generation paradigm. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_04451 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | CoT-lized Diffusion: Let's Reinforce T2I Generation Step-by-step Liu, Zheyuan Ning, Munan Zhang, Qihui Yang, Shuo Wang, Zhongrui Yang, Yiwei Xu, Xianzhe Song, Yibing Chen, Weihua Wang, Fan Yuan, Li Computer Vision and Pattern Recognition Current text-to-image (T2I) generation models struggle to align spatial composition with the input text, especially in complex scenes. Even layout-based approaches yield suboptimal spatial control, as their generation process is decoupled from layout planning, making it difficult to refine the layout during synthesis. We present CoT-Diff, a framework that brings step-by-step CoT-style reasoning into T2I generation by tightly integrating Multimodal Large Language Model (MLLM)-driven 3D layout planning with the diffusion process. CoT-Diff enables layout-aware reasoning inline within a single diffusion round: at each denoising step, the MLLM evaluates intermediate predictions, dynamically updates the 3D scene layout, and continuously guides the generation process. The updated layout is converted into semantic conditions and depth maps, which are fused into the diffusion model via a condition-aware attention mechanism, enabling precise spatial control and semantic injection. Experiments on 3D Scene benchmarks show that CoT-Diff significantly improves spatial alignment and compositional fidelity, and outperforms the state-of-the-art method by 34.7% in complex scene spatial accuracy, thereby validating the effectiveness of this entangled generation paradigm. |
| title | CoT-lized Diffusion: Let's Reinforce T2I Generation Step-by-step |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.04451 |