用单图+动作视频生成长时间可控人体视频，解决身份漂移和时序不连贯问题。
PoseGen: In-Context LoRA Finetuning for Pose-Controllable Long Human Video Generation
- 通过上下文LoRA注入外观特征，实现身份保真；同时以姿态信息控制运动细节。
- 在33小时数据上训练，生成长达数秒的视频，且身份一致性和姿态准确性超越现有方法。
- 适合需要精细人体动作控制的视频生成任务，如虚拟人、动画制作。
当前基于扩散模型的视频生成面临主体身份漂移和生成长度受限的问题。本文提出PoseGen框架，仅需一张参考图像和一段驱动视频，即可生成长时间、高一致性的人体视频。核心创新包括：在上下文内通过LoRA微调，在令牌层面注入主体外观以保持身份一致，同时在通道层面条件化姿态信息以实现精细运动控制。为突破时长限制，采用分段交错生成策略：先以共享键值缓存机制生成无重叠片段，确保背景一致性，再通过姿态感知插值完成连续拼接。尽管仅在33小时数据集上训练，其在身份保真度、姿态准确性和时序连贯性方面均优于现有最优基线。代码已开源。
原文摘要 · Abstract (English)
Generating temporally coherent, long-duration videos with precise control over subject identity and movement remains a fundamental challenge for contemporary diffusion-based models, which often suffer from identity drift and are limited to short video length. We present PoseGen, a novel framework that generates human videos of extended duration from a single reference image and a driving video. Our contributions include an in-context LoRA finetuning design that injects subject appearance at the token level for identity preservation, while simultaneously conditioning on pose information at the channel level for fine-grained motion control. To overcome duration limits, we introduce a segment-interleaved generation strategy, where non-overlapping segments are first generated with improved background consistency through a shared KV-cache mechanism, and then stitched into a continuous sequence via pose-aware interpolated generation. Despite being trained on a remarkably small 33-hour video dataset, PoseGen demonstrates superior performance over state-of-the-art baselines in identity fidelity, pose accuracy, and temporal consistency. Code is available at https://github.com/Jessie459/PoseGen .
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。