DBMSolver:面向高质量图像到图像翻译的免训练扩散桥采样器
DBMSolver: A Training-free Diffusion Bridge Sampler for High-Quality Image-to-Image Translation
把扩散桥 SDE/ODE 的半线性结构用指数积分器求解,得到免训练的一阶/二阶采样器,可直接替换现有 DBM 的采样环节,在 20 NFE 内把 FID 降低一半以上,适合需要低延迟的 I2I 翻译与修复流程。
原文依据 a training-free sampler that exploits the semi-linear structure of DBM's underlying SDE and ODE via exponential integrators
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基于扩散的图像到图像(I2I)翻译在高保真生成方面表现优异,但最优的扩散桥模型(DBM)采样缓慢,通常需要数十次函数评估(NFE)。我们提出 DBMSolver,一种免训练采样器,它通过指数积分器利用 DBM 底层 SDE 与 ODE 的半线性结构,得到高效的一阶与二阶解。这将 NFE 最多降低 5 倍,同时提升质量(例如在 DIODE 上 20 NFE 时,相比二阶基线弗雷歇初始距离下降 53%)。在最高 256×256 分辨率的图像修复、风格化和语义到图像任务上的实验表明,DBMSolver 建立了新的最优效率—质量权衡,使其具备实际应用能力。我们的代码已公开于 https://github.com/snumprlab/dbmsolver。
英文原文摘要 815 字符
Diffusion-based image-to-image (I2I) translation excels in high-fidelity generation but suffers from slow sampling in state-of-the-art Diffusion Bridge Models (DBMs), often requiring dozens of function evaluations (NFEs). We introduce DBMSolver, a training-free sampler that exploits the semi-linear structure of DBM's underlying SDE and ODE via exponential integrators, yielding highly-efficient 1st- and 2nd-order solutions. This reduces NFEs by up to 5X while boosting quality (e.g., FID drops 53% on DIODE at 20 NFEs vs. 2nd-order baseline). Experiments on inpainting, stylization, and semantics-to-image tasks across resolutions up to 256x256 show DBMSolver sets new SOTA efficiency-quality tradeoffs, enabling real-world applicability. Our code is publicly available at https://github.com/snumprlab/dbmsolver.