arXiv cs.AI· Deyuan Liu, Jungang Li, Kaiwen Zheng, Kai Jiang, Jintao Chen, Yaole Wang, Ming Lin, Tao Lin, Jianfei Chen, Jun Zhu, Jintao Zhang·· 10 小时前AI 评分29
APEX 用条件偏移实现少步流式视频生成，无需独立 fake-score 网络
Few-Step Streaming Video Generation via Condition Shifting
AI 导读
APEX 通过条件偏移在生成器内部估计生成分布的速度场，去掉 DMD 所需的独立 fake-score 网络，同时保留 S2-Avatar 的冻结参考教师、Self-Replay Forcing 与感知正则化。
正文 · 原文
Authors:Deyuan Liu, Jungang Li, Kaiwen Zheng, Kai Jiang, Jintao Chen, Yaole Wang, Ming Lin, Tao Lin, Jianfei Chen, Jun Zhu, Jintao Zhang
Abstract:Few-step sampling reduces the cost of streaming video generation. Distribution Matching Distillation (DMD) supervises few-step generators with a separate fake-score network, which adds training memory and computation and must track the evolving generator. We extend APEX to streaming video by using condition shifting to estimate the velocity field of the generated distribution within the generator. The original text condition selects generation, while a shifted condition selects fake estimation using the same parameters. This removes the independent fake network while retaining S2-Avatar's frozen reference teacher, Self-Replay Forcing, and perceptual regularization. We evaluate APEX on 180-second videos using three backbone evaluations per block and a one-step spatial Refiner, with a 70-case human study and a 320-case StreamAV-Bench comparison. Compared with S2-Avatar, APEX achieves higher video instruction fulfillment on the Progressive track (3.856 versus 3.831) and higher visual aesthetics across both tracks (0.641898 versus 0.632606). Human raters favor APEX most strongly on instruction following, with +30.0% net preference, followed by consistency and video quality at +4.3% each. Motion quality favors S2-Avatar, audiovisual synchronization is predominantly tied, and usability is unchanged. Separate measurements under matched training settings show 4.4% lower peak allocated GPU memory per GPU and 9.2% shorter mean iteration time than DMD.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.14033 [cs.CV] |
| (or arXiv:2611.14033v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.14033 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Deyuan Liu [view email]
[v1]
Tue, 10 Nov 2026 13:01:20 UTC (19,213 KB)
来源：arXiv cs.AI · arxiv.org