When to Lock Attention: Training-Free KV Control in Video Diffusion
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908877034881024 |
|---|---|
| author | Zeng, Tianyi Gao, Jincheng Wang, Tianyi Meng, Zijie Zhang, Miao Yin, Jun Sun, Haoyuan Jiao, Junfeng Claudel, Christian Tan, Junbo Wang, Xueqian |
| author_facet | Zeng, Tianyi Gao, Jincheng Wang, Tianyi Meng, Zijie Zhang, Miao Yin, Jun Sun, Haoyuan Jiao, Junfeng Claudel, Christian Tan, Junbo Wang, Xueqian |
| contents | Maintaining background consistency while enhancing foreground quality remains a core challenge in video editing. Injecting full-image information often leads to background artifacts, whereas rigid background locking severely constrains the model's capacity for foreground generation. To address this issue, we propose KV-Lock, a training-free framework tailored for DiT-based video diffusion models. Our core insight is that the hallucination metric (variance of denoising prediction) directly quantifies generation diversity, which is inherently linked to the classifier-free guidance (CFG) scale. Building upon this, KV-Lock leverages diffusion hallucination detection to dynamically schedule two key components: the fusion ratio between cached background key-values (KVs) and newly generated KVs, and the CFG scale. When hallucination risk is detected, KV-Lock strengthens background KV locking and simultaneously amplifies conditional guidance for foreground generation, thereby mitigating artifacts and improving generation fidelity. As a training-free, plug-and-play module, KV-Lock can be easily integrated into any pre-trained DiT-based models. Extensive experiments validate that our method outperforms existing approaches in improved foreground quality with high background fidelity across various video editing tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_09657 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | When to Lock Attention: Training-Free KV Control in Video Diffusion Zeng, Tianyi Gao, Jincheng Wang, Tianyi Meng, Zijie Zhang, Miao Yin, Jun Sun, Haoyuan Jiao, Junfeng Claudel, Christian Tan, Junbo Wang, Xueqian Computer Vision and Pattern Recognition Artificial Intelligence Emerging Technologies Image and Video Processing Maintaining background consistency while enhancing foreground quality remains a core challenge in video editing. Injecting full-image information often leads to background artifacts, whereas rigid background locking severely constrains the model's capacity for foreground generation. To address this issue, we propose KV-Lock, a training-free framework tailored for DiT-based video diffusion models. Our core insight is that the hallucination metric (variance of denoising prediction) directly quantifies generation diversity, which is inherently linked to the classifier-free guidance (CFG) scale. Building upon this, KV-Lock leverages diffusion hallucination detection to dynamically schedule two key components: the fusion ratio between cached background key-values (KVs) and newly generated KVs, and the CFG scale. When hallucination risk is detected, KV-Lock strengthens background KV locking and simultaneously amplifies conditional guidance for foreground generation, thereby mitigating artifacts and improving generation fidelity. As a training-free, plug-and-play module, KV-Lock can be easily integrated into any pre-trained DiT-based models. Extensive experiments validate that our method outperforms existing approaches in improved foreground quality with high background fidelity across various video editing tasks. |
| title | When to Lock Attention: Training-Free KV Control in Video Diffusion |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Emerging Technologies Image and Video Processing |
| url | https://arxiv.org/abs/2603.09657 |