Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910238735597568 |
|---|---|
| author | Jeon, Wooseok Park, Seungho Shin, Seunghyun Lee, Sangeyl Jeong, Hyeonho Jeon, Hae-Gon |
| author_facet | Jeon, Wooseok Park, Seungho Shin, Seunghyun Lee, Sangeyl Jeong, Hyeonho Jeon, Hae-Gon |
| contents | Image-to-video models often generate videos that remain overly static, compared to text-to-video models. While prior approaches mitigate this issue by weakening or modifying the image-conditioning signal, they often require additional training or sacrifice fidelity to the reference image. In this work, we identify reference-frame dominance as a key mechanism behind motion suppression. We observe that non-reference frames in I2V models allocate excessive self-attention to reference-frame key tokens, causing reference information to be over-propagated across time and suppressing inter-frame dynamics. Based on this finding, we propose DyMoS (Dynamic Motion Slider), a training-free and model-agnostic method that rebalances the attention pathway from generated frames to the reference frame during initial denoising steps. DyMoS leaves both the input image and model weights unchanged and introduces a single scalar parameter for continuous control over motion strength. Experiments across multiple state-of-the-art I2V backbones demonstrate that DyMoS consistently improves motion dynamics while maintaining visual quality and fidelity to the reference image. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_19398 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models Jeon, Wooseok Park, Seungho Shin, Seunghyun Lee, Sangeyl Jeong, Hyeonho Jeon, Hae-Gon Computer Vision and Pattern Recognition Artificial Intelligence Image-to-video models often generate videos that remain overly static, compared to text-to-video models. While prior approaches mitigate this issue by weakening or modifying the image-conditioning signal, they often require additional training or sacrifice fidelity to the reference image. In this work, we identify reference-frame dominance as a key mechanism behind motion suppression. We observe that non-reference frames in I2V models allocate excessive self-attention to reference-frame key tokens, causing reference information to be over-propagated across time and suppressing inter-frame dynamics. Based on this finding, we propose DyMoS (Dynamic Motion Slider), a training-free and model-agnostic method that rebalances the attention pathway from generated frames to the reference frame during initial denoising steps. DyMoS leaves both the input image and model weights unchanged and introduces a single scalar parameter for continuous control over motion strength. Experiments across multiple state-of-the-art I2V backbones demonstrate that DyMoS consistently improves motion dynamics while maintaining visual quality and fidelity to the reference image. |
| title | Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2605.19398 |