arXiv 2507.06119

Omni-Video: Democratizing Unified Video Understanding and Generation

By Zhiyu Tan, Hao Yang, et al.

Published 2025-07-08

Wiki summary

Explore the paper's summary, context, and related research on Papiers.

Notable breakthroughs in unified understanding and generation modeling have led to remarkable advancements in image understanding, reasoning, production and editing, yet current foundational models predominantly focus on processing images, creating a gap in the development of unified models for video understanding and generation. This report presents Omni-Video, an efficient and effective unified framework for video…

View the original paper on arXiv