Video diffusion transformers have achieved remarkable progress in high-quality video generation, but remain computationally expensive due to the quadratic complexity of attention over high-dimensional video sequences. Recent acceleration methods enhance the efficiency by exploiting the local sparsity of attention scores; yet they often struggle with accelerating the long-range computation. To address this problem, we propose VORTA, an acceleration framework with two novel components: 1) a sparse attention mechanism that efficiently captures long-range dependencies, and 2) a routing strategy that adaptively replaces full 3D attention with specialized sparse attention variants. VORTA achieves an end-to-end speedup 1.76× without loss of quality on VBench. Furthermore, it can seamlessly integrate with various other acceleration methods, such as model caching and step distillation, reaching up to speedup 14.41× with negligible performance degradation. VORTA demonstrates its efficiency and enhances the practicality of video diffusion transformers in real-world settings. Codes and weights are available at https://github.com/wenhao728/VORTA.
@article{arxiv.2505.18809,
title = {VORTA: Efficient Video Diffusion via Routing Sparse Attention},
author = {Wenhao Sun and Rong-Cheng Tu and Yifu Ding and Zhao Jin and Jingyi Liao and Shunyu Liu and Dacheng Tao},
journal= {arXiv preprint arXiv:2505.18809},
year = {2025}
}
Comments
Accepted by NeurIPS 2025. The code is available at https://github.com/wenhao728/VORTA