KeyID: Decoupled Drafting and Keyframe Editing for Identity-Preserving Video Generation
2026-08-17 • Computer Vision and Pattern Recognition
Computer Vision and Pattern RecognitionGraphicsMultimedia
AI summaryⓘ
The authors introduce KeyID, a way to make videos that keep a person's identity consistent while following text prompts. Their method separates making the motion in the video from adding the person's identity, which helps keep the identity steady over longer videos. Instead of adjusting every frame, they fix a few key frames and fill in the rest, making the process simpler and more flexible. Their approach works well for videos with multiple people and complex actions without needing extra training, and it performed strongly in a video generation challenge.
identity-preserving video generationvideo dynamicskeyframe editingmotion interpolationprompt adherenceidentity fidelitymulti-subject videosequential action generationtraining-free methodsvideo synthesis
Authors
Jianjie Luo, Yiming Zhong, Haoming Shen, Yupeng Xiao, Zhenguo Yang
Abstract
Identity-preserving video generation (IPVG) requires synthesizing videos that are faithful to both reference subjects and text prompts. Existing methods are often hindered by high tuning costs or limited input-level enhancements, struggling to maintain rigid identity consistency during complex, long-sequence actions. To address these limitations, we propose KeyID, a training-free IPVG framework that decouples the synthesis of video dynamics from the injection of identity. Specifically, KeyID comprises two components: (1) Reference-Aware Video Generation, which produces an identity-agnostic video draft aligned with multiple references, and (2) Identity-Preserved Keyframe Editing, which integrates the target identity via sparse keyframe correction and subsequent motion interpolation. By shifting from dense frame-level supervision to sparse keyframe-level refinement, KeyID effectively resolves the capacity conflict between prompt adherence and identity fidelity. Crucially, our modular design allows seamless extension to multi-subject references and complex sequential action generation without additional training. KeyID outperforms prior works and is validated by automatic and human evaluations on the official challenge benchmark, ultimately securing the runner-up position in the Track 2 (Sequential Action) of the ACM Multimedia 2026 IPVG Grand Challenge. Source code is available at https://github.com/WISLab-GDUT/KeyID.