Score: 1

Adding Additional Control to One-Step Diffusion with Joint Distribution Matching

Published: March 9, 2025 | arXiv ID: 2503.06652v2

By: Yihong Luo , Tianyang Hu , Yifan Song and more

Potential Business Impact:

Makes AI create pictures from new ideas faster.

Business Areas:
DRM Content and Publishing, Media and Entertainment, Privacy and Security

While diffusion distillation has enabled one-step generation through methods like Variational Score Distillation, adapting distilled models to emerging new controls -- such as novel structural constraints or latest user preferences -- remains challenging. Conventional approaches typically requires modifying the base diffusion model and redistilling it -- a process that is both computationally intensive and time-consuming. To address these challenges, we introduce Joint Distribution Matching (JDM), a novel approach that minimizes the reverse KL divergence between image-condition joint distributions. By deriving a tractable upper bound, JDM decouples fidelity learning from condition learning. This asymmetric distillation scheme enables our one-step student to handle controls unknown to the teacher model and facilitates improved classifier-free guidance (CFG) usage and seamless integration of human feedback learning (HFL). Experimental results demonstrate that JDM surpasses baseline methods such as multi-step ControlNet by mere one-step in most cases, while achieving state-of-the-art performance in one-step text-to-image synthesis through improved usage of CFG or HFL integration.

Page Count
11 pages

Category
Computer Science:
CV and Pattern Recognition