Salvato in:
| Autori principali: | , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2602.05551 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908919976165376 |
|---|---|
| author | Ma, Yue Wang, Zhikai Ren, Tianhao Zheng, Mingzhe Liu, Hongyu Guo, Jiayi Feng, Kunyu Xue, Yuxuan Zhao, Zixiang Schindler, Konrad Chen, Qifeng Zhang, Linfeng |
| author_facet | Ma, Yue Wang, Zhikai Ren, Tianhao Zheng, Mingzhe Liu, Hongyu Guo, Jiayi Feng, Kunyu Xue, Yuxuan Zhao, Zixiang Schindler, Konrad Chen, Qifeng Zhang, Linfeng |
| contents | Video motion transfer aims to synthesize videos by generating visual content according to a text prompt while transferring the motion pattern observed in a reference video. Recent methods predominantly use the Diffusion Transformer (DiT) architecture. To achieve satisfactory runtime, several methods attempt to accelerate the computations in the DiT, but fail to address structural sources of inefficiency. In this work, we identify and remove two types of computational redundancy in earlier work: motion redundancy arises because the generic DiT architecture does not reflect the fact that frame-to-frame motion is small and smooth; gradient redundancy occurs if one ignores that gradients change slowly along the diffusion trajectory. To mitigate motion redundancy, we mask the corresponding attention layers to a local neighborhood such that interaction weights are not computed unnecessarily distant image regions. To exploit gradient redundancy, we design an optimization scheme that reuses gradients from previous diffusion steps and skips unwarranted gradient computations. On average, FastVMT achieves a 3.43x speedup without degrading the visual fidelity or the temporal consistency of the generated videos. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_05551 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | FastVMT: Eliminating Redundancy in Video Motion Transfer Ma, Yue Wang, Zhikai Ren, Tianhao Zheng, Mingzhe Liu, Hongyu Guo, Jiayi Feng, Kunyu Xue, Yuxuan Zhao, Zixiang Schindler, Konrad Chen, Qifeng Zhang, Linfeng Computer Vision and Pattern Recognition Video motion transfer aims to synthesize videos by generating visual content according to a text prompt while transferring the motion pattern observed in a reference video. Recent methods predominantly use the Diffusion Transformer (DiT) architecture. To achieve satisfactory runtime, several methods attempt to accelerate the computations in the DiT, but fail to address structural sources of inefficiency. In this work, we identify and remove two types of computational redundancy in earlier work: motion redundancy arises because the generic DiT architecture does not reflect the fact that frame-to-frame motion is small and smooth; gradient redundancy occurs if one ignores that gradients change slowly along the diffusion trajectory. To mitigate motion redundancy, we mask the corresponding attention layers to a local neighborhood such that interaction weights are not computed unnecessarily distant image regions. To exploit gradient redundancy, we design an optimization scheme that reuses gradients from previous diffusion steps and skips unwarranted gradient computations. On average, FastVMT achieves a 3.43x speedup without degrading the visual fidelity or the temporal consistency of the generated videos. |
| title | FastVMT: Eliminating Redundancy in Video Motion Transfer |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.05551 |