Sora Faces New Competition from Shengshu AI in China with Launch of Text-to-Video Tool
Chinese startup Shengshu AI released Vidu, a text-to-video application that supports both Chinese and English language prompts, in an attempt to compete with OpenAI's Sora. Vidu, which is available via its website, allows users to make four or eight-second movies and joins other Chinese start-ups such as Zhipu AI and Kuaishou Technology in providing text-to-video services. Shengshu's Vidu, built using the U-ViT architecture, can make four-second clips in 30 seconds, increasing its commercial viability in the animation and video industries.
Chinese startup Shengshu AI has made its text-to-video solution, Vidu, available to a global audience, with support for both Chinese and English text prompts. This move reflects China's tech industry's drive to rival OpenAI's Sora.
Accessible via its official website, Vidu joins other Chinese start-ups like Zhipu AI and Kuaishou Technology in offering text-to-video services. Users can create clips of four or eight seconds upon registration.
Shengshu unveiled Vidu in April, shortly after OpenAI's Sora debt, positioning itself as the first Chinese company to compete directly with Sora. The tool can generate a four-second clip in just 30 seconds, making it one of the fastest options available.
Vidu is powered by Shengshu's self-developed architecture, U-ViT, which is described in a research article conducted by Zhu Jun, the company's principal scientist and a computer science professor at Tsinghua University. Bao Fan, another Tsinghua author, is Shengshu's chief technology officer.
Tang Jiayu, Shengshu's CEO and a Tsinghua graduate, showed confidence in catching up with Sora compared to OpenAI's GPT-4. Vidu now features a character-to-video function, which increases its commercial viability in the animation and entertainment businesses.
Zhang Xudong, Shengshu AI's product director, anticipates people submitting photographs of actual or animated characters and then utilising language instructions to bring them to life. The goal is to allow several characters to interact in scenes, similar to film production, by combining AI tools with traditional methods, according to the principal scientist and computer science professor at Tsinghua University.
Chinese start-up Shengshu AI launched Vidu, a text-to-video tool supporting Chinese and English text prompts, aiming to rival OpenAI's Sora.
Vidu, accessible via its website, allows users to create four or eight-second clips and joins other Chinese start-ups like Zhipu AI and Kuaishou Technology in offering text-to-video services.
Shengshu's Vidu, developed with the U-ViT architecture, can generate four-second clips in 30 seconds, expanding its potential for commercial use in animation and content industries.
Source: SCMP