China's startup launches text-to-video generator to take on OpenAI's Sora
A Chinese startup has unveiled Vidu, an innovative artificial intelligence (AI) tool capable of transforming text prompts into short videos. Developed by Shengshu Technology in collaboration with Tsinghua University, the tool is similar to OpenAI's Sora. However, Vidu's video generation capacity is restricted to 16 seconds while its popular rival Sora can generate 60-second-long clips. The unique feature of this Chinese AI tool is its ability to generate 1080p resolution videos from simple text prompts.
Vidu is touted as a product of self-reliant innovation
Zhu Jun, the chief scientist at Shengshu and deputy dean at Tsinghua's Institute for AI, introduced Vidu at the Zhongguancun Forum in Beijing. Zhu hailed Vidu as a product of self-reliant innovation with numerous breakthroughs. He described it as an "imaginative" tool capable of simulating the physical world. During its launch, the startup showcased Vidu's capabilities through several clips, including a panda playing a guitar on grass and a puppy swimming in a pool.
Vidu's launch sparks optimism in China's AI sector
The introduction of Vidu has sparked optimism within China as it aims to compete with global leaders in generative AI technology such as Microsoft-backed OpenAI. However, Chinese tech companies have shown less enthusiasm for Sora compared to OpenAI's ChatGPT, the popular AI chatbot. Industry insiders suggest this could be due to inadequate computing power. According to Beijing-based technical consultant Li Yangwei, Sora requires eight NVIDIA A100 graphics processing units (GPUs) operating for over three hours to create a one-minute clip.
