Classifier-free guidance (CFG) is crucial for improving both generation quality and alignment between the input condition and final output in diffusion models. While a high guidance scale is generally required to enhance these aspects, it also causes oversaturation and unrealistic artifacts. In this paper, we revisit the CFG update rule and introduce modifications to address this issue. We first decompose the update term in CFG into parallel and orthogonal components with respect to the conditional model prediction and observe that the parallel component primarily causes oversaturation, while the orthogonal component enhances image quality. Accordingly, we propose down-weighting the parallel component to achieve high-quality generations without oversaturation. Additionally, we draw a connection between CFG and gradient ascent and introduce a new rescaling and momentum method for the CFG update rule based on this insight. Our approach, termed adaptive projected guidance (APG), retains the quality-boosting advantages of CFG while enabling the use of higher guidance scales without oversaturation. APG is easy to implement and introduces practically no additional computational overhead to the sampling process. Through extensive experiments, we demonstrate that APG is compatible with various conditional diffusion models and samplers, leading to improved FID, recall, and saturation scores while maintaining precision comparable to CFG, making our method a superior plug-and-play alternative to standard classifier-free guidance.
Physically-Grounded Manifold Projection Model for Generalizable Metal Artifact Reduction in Dental CBCT
School of Cyberspace, Hangzhou Dianzi University（电子科技大学信息学院）；Innovation Center for Electronic Design Automation Technology, Hangzhou Dianzi University（电子科技大学电子设计自动化创新中心）；Hangzhou Geriatric Stomatology Hospital, Hangzhou Dental Hospital Group（杭州老年口腔医院，杭州口腔医院集团）；School of Computing and Mathematical Sciences, University of Leicester（莱斯特大学计算与数学科学学院）
A Conditional Denoising Diffusion Probabilistic Model for Radio Interferometric Image Reconstruction
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Hong Kong University of Science and Technology（香港科技大学）；Guangzhou University（广州大学）
Guidance Matters: Rethinking the Evaluation Pitfall for Text-to-Image Generation
The Hong Kong University of Science and Technology (Guangzhou)（香港科学与技术大学（广州））；Harbin Institute of Technology (Shenzhen)（哈尔滨工业大学（深圳））；Cogniser Information Technology（Cogniser信息科技）
Sparse-to-Complete: From Sparse Image Captures to Complete 3D Scenes
State Key Lab of CAD\&CG, Zhejiang University（浙江大学CAD与CG国家重点实验室）；University of Utah（犹他大学）；Hangzhou Research Institute of Holographic and AI Technology（杭州全息与人工智能技术研究院）
Backend-Agnostic Sparse Attention for Fast High-Resolution Visual Generation
School of Artificial Intelligence, Shanghai Jiao Tong University（上海交通大学人工智能学院）；School of Data Science, Fudan University（复旦大学数据科学学院）；School of Computing and Data Science, The University of Hong Kong（香港大学计算及数据科学学院）
针对扩散变换器高分辨率生成中全注意力计算昂贵的问题，提出后端无关的稀疏注意力 BASA，通过跨块窗口移位实现全局交互，无需定制内核，在 FLUX 和 Wan 上实现接近理论值的加速并保持生成质量。
PANDORA: Pixel-wise Attention Dissolution and Latent Guidance for Zero-Shot Object Removal
University of Science, Ho Chi Minh City, Vietnam（越南胡志明市理科大学）；Vietnam National University, Ho Chi Minh City, Vietnam（越南胡志明市国家大学）；University of Dayton, Ohio, United States（美国俄亥俄州代顿大学）
