自适应视频蒸馏：缓解少步生成中的过饱和与时序坍缩
Adaptive Video Distillation: Mitigating Oversaturation and Temporal Collapse in Few-Step Generation
针对视频扩散模型蒸馏中的过饱和与时序坍缩问题，提出自适应回归损失和时序正则化损失，可直接用于改进少步视频生成模型的训练稳定性与运动真实性。
原文依据 an adaptive regression loss that dynamically adjusts spatial supervision weights to prevent artifacts arising from excessive distribution shifts
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视频生成近来已成为生成式人工智能领域的核心任务。然而，视频合成固有的巨大计算成本使得模型蒸馏成为高效部署的关键技术。尽管意义重大，专门针对视频扩散模型设计的方法仍然稀缺。现有方法往往直接套用图像蒸馏技术，这常常导致过饱和、时序不一致和模式坍缩等伪影。为应对这些挑战，我们提出了一种专为视频扩散模型定制的新型蒸馏框架。其核心创新包括：（1）自适应回归损失，动态调整空间监督权重，以防止因过度分布偏移而产生的伪影；（2）时序正则化损失，用以抵消时序坍缩，促进平滑且物理合理的采样轨迹；（3）推理时帧插值策略，在保持感知质量的同时降低采样开销。在 VBench 和 VBench2 基准上的大量实验与消融研究表明，我们的方法实现了稳定的少步视频合成，显著提升了感知保真度和运动真实感。在多项指标上，该方法持续优于现有的蒸馏基线。
英文原文摘要 1369 字符
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.