China’s first Sora-level text-to-video large model Vidu unveiled
At the 2024 Zhongguancun Forum in Beijing, a groundbreaking revelation in artificial intelligence captured global attention with the introduction of Vidu, China’s inaugural Sora-level text-to-video large model. Developed collaboratively by Chinese AI firm Shengshu Technology and Tsinghua University, Vidu marks a significant leap forward in AI innovation, amplifying competition on the global stage.
According to reports from China Daily, Vidu showcases remarkable capabilities by seamlessly generating high-definition videos, each 16 seconds long and in 1080p resolution, with just a single click. This unprecedented feat underscores Vidu’s extended duration, exceptional consistency, and dynamic functionalities, positioning it on par with the esteemed Sora model.
Sora, an AI marvel engineered by the United States’ OpenAI earlier in the year, has garnered widespread acclaim for its ability to translate text instructions into immersive and imaginative visual sequences, revolutionizing the tech landscape. Vidu, however, distinguishes itself by incorporating Chinese cultural elements, including iconic symbols like the panda and dragon, into its generated content.
Notably, Vidu’s core architecture was initiated in September 2022, predating Sora’s architectural developments. This early groundwork underscores China’s proactive approach to AI research and development, solidifying its status as a formidable player in the global AI arena.
As Vidu emerges onto the scene, it heralds a new era of creative possibilities in text-to-video generation, providing creators with unprecedented tools to manifest their visions digitally. With its groundbreaking capabilities and cultural sensitivity, Vidu is poised to shape the future of artificial intelligence, paving the way for innovative applications across industries.
By Cheng Yu | chinadaily.com.cn | Updated: 2024-04-27 14:04