通过感知残差幅度变化，实现视频生成的高效自适应跳步加速。
MagCache: Fast Video Generation with Magnitude-Aware Cache
- 基于残差幅度单调下降规律，动态判断可跳过的时间步。
- 单样本校准即可实现2.10倍至2.68倍加速，保持高画质。
- 适用于多种模型，无需针对特定提示词调参，鲁棒性强。
现有视频扩散模型加速方法多依赖统一启发式或时序嵌入变体来跳过时间步并复用缓存特征，通常需大量定制化提示词进行校准，且易因提示词特异性过拟合导致输出不一致。本文发现不同模型与提示词下均存在统一的幅度规律：连续残差输出的幅度比呈单调递减，在多数时间步平稳下降，最后几步快速衰减。基于此，提出幅度感知缓存（MagCache），通过误差建模机制与自适应缓存策略，动态跳过低贡献时间步。相比需数十个样本校准的方法，仅需单样本即可完成校准。实验表明，MagCache在Open-Sora、CogVideoX、Wan 2.1和HunyuanVideo上实现2.10x–2.68x加速，同时在相同计算预算下显著优于现有方法，在LPIPS、SSIM和PSNR指标上表现更优。
原文摘要 · Abstract (English)
Existing acceleration techniques for video diffusion models often rely on uniform heuristics or time-embedding variants to skip timesteps and reuse cached features. These approaches typically require extensive calibration with curated prompts and risk inconsistent outputs due to prompt-specific overfitting. In this paper, we introduce a novel and robust discovery: a unified magnitude law observed across different models and prompts. Specifically, the magnitude ratio of successive residual outputs decreases monotonically, steadily in most timesteps while rapidly in the last several steps. Leveraging this insight, we introduce a Magnitude-aware Cache (MagCache) that adaptively skips unimportant timesteps using an error modeling mechanism and adaptive caching strategy. Unlike existing methods requiring dozens of curated samples for calibration, MagCache only requires a single sample for calibration. Experimental results show that MagCache achieves 2.10x-2.68x speedups on Open-Sora, CogVideoX, Wan 2.1, and HunyuanVideo, while preserving superior visual fidelity. It significantly outperforms existing methods in LPIPS, SSIM, and PSNR, under similar computational budgets.
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。