针对视频生成模型蒸馏中的过饱和与时间崩溃问题，提出自适应蒸馏框架。
Adaptive Video Distillation: Mitigating Oversaturation and Temporal Collapse in Few-Step Generation
- 设计动态空间监督权重的回归损失，缓解分布偏移导致的失真。
- 引入时间正则化损失，提升采样轨迹的平滑性与物理合理性。
- 推理时帧插值降低计算开销，适合快速生成高质量视频。
视频生成已成为生成式AI的核心任务，但其高昂的计算成本使得模型蒸馏成为高效部署的关键。然而，现有方法多直接沿用图像蒸馏技术，常引发过饱和、时间不一致和模式坍塌等缺陷。为此，本文提出专为视频扩散模型设计的新型蒸馏框架：（1）自适应回归损失，动态调整空间监督权重以抑制分布偏移带来的伪影；（2）时间正则化损失，对抗时间坍缩，促进平滑且物理合理的采样路径；（3）推理阶段帧插值策略，在保持感知质量的同时降低采样开销。在VBench与VBench2基准上的大量实验与消融研究显示，该方法实现了稳定的少步视频生成，显著提升感知保真度与运动真实性，全面优于现有蒸馏基线。
原文摘要 · Abstract (English)
Video generation has recently emerged as a central task in the field of generative AI. However, the substantial computational cost inherent in video synthesis makes model distillation a critical technique for efficient deployment. Despite its significance, there is a scarcity of methods specifically designed for video diffusion models. Prevailing approaches often directly adapt image distillation techniques, which frequently lead to artifacts such as oversaturation, temporal inconsistency, and mode collapse. To address these challenges, we propose a novel distillation framework tailored specifically for video diffusion models. Its core innovations include: (1) an adaptive regression loss that dynamically adjusts spatial supervision weights to prevent artifacts arising from excessive distribution shifts; (2) a temporal regularization loss to counteract temporal collapse, promoting smooth and physically plausible sampling trajectories; and (3) an inference-time frame interpolation strategy that reduces sampling overhead while preserving perceptual quality. Extensive experiments and ablation studies on the VBench and VBench2 benchmarks demonstrate that our method achieves stable few-step video synthesis, significantly enhancing perceptual fidelity and motion realism. It consistently outperforms existing distillation baselines across multiple metrics.
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。