Jul 2026
Keyframe-Anchored Identity Preservation for Sequential-Action Video Generation
This work presents a training-free, three-stage pipeline framework that produces a keyframe sequence by conditioning each frame jointly on the reference identity and its predecessor, thereby decoupling time-invariant appearance from time-varying pose.
Zhenjie Liu, Bin-Yan Chen, Hao Chen et al.
· arXiv.org · 0 citations