对话生数科技创始人兼首席科学家朱军：AI视频生成正迈入“高可控”时代

Core Insights - The development of large models, particularly in AI video generation, is rapidly evolving, with significant advancements expected in 2025 [1][3] Company Insights - Shengshu Technology has officially launched the Vidu Q1, the first high-controllability video large model in the industry, which is set to go global in April [1][3] - Vidu Q1 has achieved major technical breakthroughs, allowing for spatial layout information as input, significantly enhancing video generation controllability [3][4] - The company’s SaaS products have reached over 10 million users within 100 days of launch, marking the fastest growth globally [4] - Shengshu Technology is focusing on multi-modal large model development, with video being one of the expressions of this technology [4][6] - The company emphasizes the importance of continuous innovation and the ability to meet user demands in different stages of development [4][7] Industry Insights - The commercialization path for video large models is expected to be more diverse compared to language models, with a faster market acceptance due to broader application scenarios [6][7] - The video industry is unlikely to see a dominant player like DeepSeek in language models, as the competition is more diversified and less crowded [7][8] - The industry is moving towards longer, more narrative-driven video content, transitioning from short videos to more complex storytelling [8]