Reflection-Guided Dynamics Rectification for Text-to-Image Generation

Key points

• Fine-grained rectification of existing generations under a limited additional budget remains a key problem in test-time optimization (TTO) for image generation.1
• Moreover, RGDR generates flexible candidates while adjudicating between them and the reference.1
• As experiments show, rectification from RGDR starts from the existing base trajectory and reflection, and RGDR is substantially more effective than former trajectory exploration.1
• Under matched budgets for additional image generation, RGDR's gain in the cross-model average final checklist score over the reference is $1.51\times$ that of TIR prompt refinement and $1.81\times$ that of Dynamic CFG.1

Sources (1)

• [1]Reflection-Guided Dynamics Rectification for Text-to-Image Generation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 18, 09:47 AM

  To narrow it, we propose Reflection-Guided Dynamics Rectification (RGDR), which turns reflections on completed outputs into corrective interventions on the process of sampling dynamics at the trajectory level.

  Fine-grained rectification of existing generations under a limited additional budget remains a key problem in test-time optimization (TTO) for image generation.

Extractive summary: sentences quoted from the sources.