ByteDance, TikTok's parent company, has launched a new app called Jimeng AI in China that has the ability to generate videos based on text prompts -- similar to OpenAI’s recently announced Sora and a number of projects led by Chinese tech firms.
If integrated, the app could have a significant effect on TikTok's creation tools.
According to Jimeng AI’s website, the app invites users to create video clips up to five seconds long as well as images, by entering a written description of what they want to see.
Examples included on the website's main page show a zebra underwater with sunglasses on, a dog swimming, lights reflecting off an astronaut's mirrored helmet, and the very real face of an AI-generated child.
The results of Jimeng AI resemble Sora, a text-to-video model unveiled by OpenAI earlier this year, which is not yet open to public use but can also create 1080p videos up to one minute in length -– far longer than any other AI video-generator on the market.
advertisement
advertisement
Sora has the ability to generate looping videos, edit existing videos and simulate digital worlds. Described by senior Nvidia researcher Jim Fan as a “data-driven physics engine,” Sora can also replicate video-game environments like Minecraft while controlling the player character.
According to a report in Reuters, Chinese companies have been rapidly developing tools similar to Jimeng AI -- with several launching models accessible to users, including Kuaishou, which opened its Kling AI text-to-video model to a global audience last month.
However, because of its owner, Jimeng AI is in the position to make a future impact on the social-media space. ByteDance’s massive video-sharing platform TikTok -- and its Chinese counterpart Douyin -- could see the integration of this text-to-video technology, with new video-first creation tools.
Despite a potential nationwide ban in the U.S., TikTok is steadfast in its goal of incorporating an array of generative AI features into the app, including chatbots, search tools, music generation tools and profile picture simulators.
Text-to-video would align directly with the app's original video-first identity, expanding how users can share content and trends on the app, and inviting brands to showcase their products in more immersive, fantastical ways via influencer content.
Following its initial Android release on July 31, Jimeng AI -- developed by ByteDance-owned Faceu Technology -- is now available on the Apple App Store for Chinese users.