用语义退化条件引导扩散模型
Guiding Diffusion Models with Semantically Degraded Conditions
CDG 提出用语义退化条件替代空提示词作为负样本，为扩散模型引导提供了一种无需训练、即插即用的改进方案，可直接用于提升文本到图像生成的组合准确性和图文对齐。
原文依据 We propose Condition-Degradation Guidance (CDG), a novel paradigm that replaces the null prompt with a strategically degraded condition, c_deg.
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论无分类器引导（CFG）是现代文本到图像模型的基石，然而它依赖语义空洞的空提示词，生成的引导信号容易产生几何纠缠。这是限制其精度的关键因素，导致在复杂组合任务中出现已记录的失败。我们提出条件退化引导（CDG），一种新颖的范式，用策略性退化的条件 c_deg 替代空提示词。这将引导从粗略的“好 vs. 空”对比重新定义为更精细的“好 vs. 几乎好”判别，从而迫使模型捕捉细粒度的语义区分。我们发现 Transformer 文本编码器中的 token 分为两种功能角色：编码对象语义的内容 token 和捕捉全局上下文的上下文聚合 token。通过仅选择性退化前者，CDG 无需外部模型或训练即可构建 c_deg。在包括 Stable Diffusion 3、FLUX 和 Qwen-Image 在内的多种架构上验证，CDG 显著提升了组合准确性和图文对齐。作为一个轻量级、即插即用的模块，它以可忽略的计算开销实现了这一点。我们的工作挑战了对静态、信息稀疏负样本的依赖，并确立了扩散引导的新原则：构建自适应、语义感知的负样本对于实现精确的语义控制至关重要。代码见 https://github.com/Ming-321/Classifier-Degradation-Guidance。
英文原文摘要 1522 字符
Classifier-Free Guidance (CFG) is a cornerstone of modern text-to-image models, yet its reliance on a semantically vacuous null prompt generates a guidance signal prone to geometric entanglement. This is a key factor limiting its precision, leading to well-documented failures in complex compositional tasks. We propose Condition-Degradation Guidance (CDG), a novel paradigm that replaces the null prompt with a strategically degraded condition, c_deg. This reframes guidance from a coarse "good vs. null" contrast to a more refined "good vs. almost good" discrimination, thereby compelling the model to capture fine-grained semantic distinctions. We find that tokens in transformer text encoders split into two functional roles: content tokens encoding object semantics, and context-aggregating tokens capturing global context. By selectively degrading only the former, CDG constructs c_deg without external models or training. Validated across diverse architectures including Stable Diffusion 3, FLUX, and Qwen-Image, CDG markedly improves compositional accuracy and text-image alignment. As a lightweight, plug-and-play module, it achieves this with negligible computational overhead. Our work challenges the reliance on static, information-sparse negative samples and establishes a new principle for diffusion guidance: the construction of adaptive, semantically aware negative samples is critical to achieving precise semantic control. Code is available at https://github.com/Ming-321/Classifier-Degradation-Guidance.