Lune

CVPR2026Top-tier venue

Synthetic Curriculum Reinforces Compositional Text-to-Image Generation

Shijian Wang, Runhao Fu, Siyi Zhao, Qingqin Zhan, Xingjian Wang, Jiarui Jin, Yuan Lu, Hanqian Wu, Cunjian Chen

Abstract

Text-to-Image (T2I) generation has long been an open problem, with compositional synthesis remaining particularly challenging. This task requires accurate rendering of complex scenes containing multiple objects that exhibit diverse attributes as well as intricate spatial and semantic relationships, demanding both precise object placement and coherent inter-object interactions. In this paper, we propose a novel compositional curriculum reinforcement learning framework named CompGen that addresses compositional weakness in existing T2I models. Specifically, we leverage scene graphs to establish a novel difficulty criterion for compositional ability and develop a corresponding adaptive Markov Chain Monte Carlo graph sampling algorithm. This difficulty-aware approach enables the synthesis of training curriculum data that progressively optimize T2I models through reinforcement learning. We integrate our curriculum learning approach into Group Relative Policy Optimization (GRPO) and investigate different curriculum scheduling strategies. Our experiments reveal that CompGen exhibits distinct scaling curves under different curriculum scheduling strategies, with easy-to-hard and Gaussian sampling strategies yielding superior scaling performance compared to random sampling. Extensive experiments demonstrate that CompGen significantly enhances compositional generation capabilities for both diffusion-based and auto-regressive T2I models, highlighting its effectiveness in improving the compositional T2I generation systems.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

1 top-tier paper has cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Visual Instruction Tuning cites High-Resolution Image Synthesis with Latent Diffusion Models
  • Visual Instruction Tuning cites Learning Transferable Visual Models From Natural Language Supervision
  • Visual Instruction Tuning cites Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
  • BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation cites Learning Transferable Visual Models From Natural Language Supervision
  • Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding cites High-Resolution Image Synthesis with Latent Diffusion Models
  • Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding cites Learning Transferable Visual Models From Natural Language Supervision
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers1

  1. DeepAgent: A General Reasoning Agent with Scalable ToolsetsXiaoxi Li, Wenxiang Jiao, Jiarui Jin, Guanting Dong et al.WWW 2026 · 38 citations

Builds on35

  1. Learning Transferable Visual Models From Natural Language SupervisionAlec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh et al.ICML 2021 · 47,906 citations
  2. High-Resolution Image Synthesis with Latent Diffusion ModelsRobin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser et al.CVPR 2022 · 13,123 citations
  3. Visual Instruction TuningHaotian Liu, Chunyuan Li, Qingyang Wu, Yong Jae LeeNeurIPS 2023 · 11,349 citations
  4. Photorealistic Text-to-Image Diffusion Models with Deep Language UnderstandingChitwan Saharia, William Chan, Saurabh Saxena, Lala Li et al.NeurIPS 2022 · 8,965 citations
  5. BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and GenerationJunnan Li, Dongxu Li, Caiming Xiong, Steven C. H. HoiICML 2022 · 6,549 citations

Related papers

  1. Curriculum Group Policy Optimization: Adaptive Sampling for Unleashing the Potential of Text-to-Image GenerationBaoteng Li, Xianghao Zang, Xinran Wang, Xiangyu Na et al.CVPR 2026
  2. HiCoGen: Hierarchical Compositional Text-to-Image Generation in Diffusion Models via Reinforcement LearningHongji Yang, Yucheng Zhou, Wencheng Han, Runzhou Tao et al.CVPR 2026 · 4 citations
  3. Reinforcement Learning Meets Masked Generative Models: Mask-GRPO for Text-to-Image GenerationYifu Luo, Xinhao Hu, Keyu Fan, Haoyuan Sun et al.NeurIPS 2025 · 12 citations
  4. Curriculum Direct Preference Optimization for Diffusion and Consistency ModelsFlorinel-Alin Croitoru, Vlad Hondru, Radu Tudor Ionescu, Nicu Sebe et al.CVPR 2025
  5. Environment Generation for Zero-Shot Compositional Reinforcement LearningIzzeddin Gur, Natasha Jaques, Yingjie Miao, Jongwook Choi et al.NeurIPS 2021 · 50 citations
Dusk over the sea between two cliffs drawn in fine vertical lines