(Yicai) Feb. 21 -- The text-to-video artificial intelligence generator that ByteDance is developing is way behind overseas tools for definition, accuracy, and video length, a source at the Chinese internet giant is reported to have said.

Boximator, a tool similar to OpenAI's Sora, is not yet available as a complete product, The Paper reported yesterday, citing a researcher at Beijing-based ByteDance. 

It can accurately control the movements of people and objects in a video via text inputs unlike mainstream AI video tools such as Gen-2 and Pika from California-based startups Runway and Pika Labs, the person added.

OpenAI released Sora on Feb. 15, with the tool attracting huge market attention for its ability to generate realistic 60-second videos in high definition based on short text instructions. 

Boximator is still in development, and a website for testing will be available in two to three months, according to ByteDance engineers.

Many industry professionals believe ByteDance's all-in-one video editor and graphic design tool CapCut can catch up with Sora, but the Chinese firm has not said whether Boximator will be integrated into CapCut. 

On Feb. 7, Kelly Zhang stepped down as chief executive of Douyin Group, the unit of ByteDance that runs short-video platform Douyin, the Chinese version of TikTok, and said she will put all her energy into CapCut. CapCut had over 100 monthly active users and was China's biggest mobile video editing tool as of 2021.

Editor: Martin Kadiev