不重训不降质，让图像生成轨迹自发扩散以提升多样性。
Letting Trajectories Spread: Quality-Preserving Control for Diverse Flow Matching
- 通过特征空间引导与时间调度扰动，实现生成轨迹横向分散。
- 在固定采样次数下，Vendi Score 和 Brisque 指标显著优于基线。
- 适合追求高多样性且不牺牲图像质量的文本生成应用。
基于流的文生图模型遵循确定性轨迹，导致在有限采样预算下难以探索多样模式。现有提升多样性的方法常依赖重训练或损害图像保真度。为此，我们提出一种无需训练、仅在推理时使用的控制机制，使流本身具备多样性感知能力。核心思想是通过几何上与质量优化方向解耦的引导来促进多样性。方法同时利用特征空间目标鼓励轨迹横向扩展，并通过时间调度的随机扰动重新引入不确定性。关键在于，该扰动被投影为与生成流正交，从而在不破坏图像细节或提示一致性的情况下提升变化性。理论上，该设计单调增加体积代理指标，近似保持边缘分布，为生成质量的鲁棒性提供理论解释。实验表明，在多个文生图设置下，固定采样预算内，本方法在 Vendi Score 与 Brisque 指标上持续超越强基线，同时维持高质量与提示对齐。
原文摘要 · Abstract (English)
Flow-based text-to-image models follow deterministic trajectories, making it costly to explore diverse modes under limited sampling budgets. Existing approaches to improving diversity often rely on retraining or degrade image fidelity. To address this limitation, we present a training-free, inference-time control mechanism that makes the flow itself diversity-aware. Our core insight is to encourage diversity through guidance that is geometrically decoupled from the mode's quality-seeking direction. Our method simultaneously encourages lateral spread among trajectories via a feature-space objective and reintroduces uncertainty through a time-scheduled stochastic perturbation. Crucially, this perturbation is projected to be orthogonal to the generation flow, a geometric constraint that allows it to boost variation without degrading image details or prompt fidelity. Theoretically, we show that this design monotonically increases a volume surrogate while approximately preserving the marginal distribution, providing a principled explanation for the robustness of generation quality. Empirically, across multiple text-to-image settings under fixed sampling budgets, our method consistently improves diversity metrics such as the Vendi Score and Brisque over strong baselines, while upholding image quality and alignment.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。