Salvato in:
| Autori principali: | , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2507.12952 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912488606400512 |
|---|---|
| author | Jiang, Jiaxiu Li, Wenbo Ren, Jingjing Qiu, Yuping Guo, Yong Xu, Xiaogang Wu, Han Zuo, Wangmeng |
| author_facet | Jiang, Jiaxiu Li, Wenbo Ren, Jingjing Qiu, Yuping Guo, Yong Xu, Xiaogang Wu, Han Zuo, Wangmeng |
| contents | Despite recent advances in diffusion transformers (DiTs) for text-to-video generation, scaling to long-duration content remains challenging due to the quadratic complexity of self-attention. While prior efforts -- such as sparse attention and temporally autoregressive models -- offer partial relief, they often compromise temporal coherence or scalability. We introduce LoViC, a DiT-based framework trained on million-scale open-domain videos, designed to produce long, coherent videos through a segment-wise generation process. At the core of our approach is FlexFormer, an expressive autoencoder that jointly compresses video and text into unified latent representations. It supports variable-length inputs with linearly adjustable compression rates, enabled by a single query token design based on the Q-Former architecture. Additionally, by encoding temporal context through position-aware mechanisms, our model seamlessly supports prediction, retradiction, interpolation, and multi-shot generation within a unified paradigm. Extensive experiments across diverse tasks validate the effectiveness and versatility of our approach. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_12952 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LoViC: Efficient Long Video Generation with Context Compression Jiang, Jiaxiu Li, Wenbo Ren, Jingjing Qiu, Yuping Guo, Yong Xu, Xiaogang Wu, Han Zuo, Wangmeng Computer Vision and Pattern Recognition Despite recent advances in diffusion transformers (DiTs) for text-to-video generation, scaling to long-duration content remains challenging due to the quadratic complexity of self-attention. While prior efforts -- such as sparse attention and temporally autoregressive models -- offer partial relief, they often compromise temporal coherence or scalability. We introduce LoViC, a DiT-based framework trained on million-scale open-domain videos, designed to produce long, coherent videos through a segment-wise generation process. At the core of our approach is FlexFormer, an expressive autoencoder that jointly compresses video and text into unified latent representations. It supports variable-length inputs with linearly adjustable compression rates, enabled by a single query token design based on the Q-Former architecture. Additionally, by encoding temporal context through position-aware mechanisms, our model seamlessly supports prediction, retradiction, interpolation, and multi-shot generation within a unified paradigm. Extensive experiments across diverse tasks validate the effectiveness and versatility of our approach. |
| title | LoViC: Efficient Long Video Generation with Context Compression |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2507.12952 |