Mastering Text-to-image Diffusion: Recaptioning, Planning, And Generating With Multimodal Llms
Ling Yang, Zhaochen Yu, Chenlin Meng, Minkai Xu, Stefano Ermon, Bin Cui . No Venue 2024
[Code]
[Paper]
Diffusion models have exhibit exceptional performance in text-to-image generation and editing. However, existing methods often face challenges when handling complex text prompts that involve multiple objects with multiple attributes and relationships. In this paper, we propose a brand new training-free text-to-image generation/editing framework, namely Recaption, Plan and Generate (RPG), harnessing the powerful chain-of-thought reasoning ability of multimodal LLMs to enhance the compositionality of text-to-image diffusion models. Our approach employs the MLLM as a global planner to decompose the process of generating complex images into multiple simpler generation tasks within subregions. We propose complementary regional diffusion to enable region-wise compositional generation. Furthermore, we integrate text-guided image generation and editing within the proposed RPG in a closed-loop fashion, thereby enhancing generalization ability. Extensive experiments demonstrate our RPG outperforms state-of-the-art text-to-image diffusion models, including DALL-E 3 and SDXL, particularly in multi-category object composition and text-image semantic alignment. Notably, our RPG framework exhibits wide compatibility with various MLLM architectures (e.g., MiniGPT-4) and diffusion backbones (e.g., ControlNet). Our code is available at: https://github.com/YangLing0818/RPG-DiffusionMaster