After fake texts created with AI and images that look real but are not, it is now the turn of videos; and this could be a bigger revolution than ChatGPT and Dall-E were in their time.
We’ve already seen a demonstration of what’s possible with today’s generative AI technology, with the introduction of videos created with Sora earlier this year; this app developed by OpenAI is capable of creating professional-grade videos simply by indicating what you want. It’s as dangerous as it sounds, which is why Sora is not yet available to the public and is reserved for only a few users who collaborate with the company. But not all companies are so reluctant, and some like ByteDance are taking advantage of OpenAI’s doubts to launch their own video creation apps.
ByteDance is the creator of TikTok, the most popular app among young people and which could be banned in the United States. One of its startups, Faceu Technology, presented this week Jimeng IAa new app launched in the Android and iOS app stores for the Chinese market; and what has already been seen has a lot of potential, for better and for worse.
Jimeng AI is able to convert any text into a short video that the user can then edit and publish on social media, including TikTok (known in China as Douyin).
This is not the only app available in China that promises to do the same; the big difference is that Jimeng AI is not free, but requires a payment of 69 yuan (8.80 dollars at the exchange rate) per monthor 659 yuan (84.30 dollars) to use it. In exchange, paying users can create 168 videos per month, in addition to 2,050 images (since the app also has generative AI for images).
Douyin cut and published “Jimeng AI”, which surpassed “Kuaishou Keling” according to the data of the monthly AI product list. I tried the product and used the given 60 points to generate a 12-second standard video.
Quick note: “Obsidian is an AI editor that wears a flight suit and swims in the sea” https://t.co/ilOKd9FxGN
Below is the generated video 👇, it looks horrible 😂 It looks like it’s based on delivering goods pic.twitter.com/VwooG8aNur– Tom Huang (@tuturetom) July 8, 2024
The first videos created with Jimeng AI are already circulating on Chinese social networks like Douyin and Bilibili, and some have made the leap to apps like X (formerly Twitter). The results are surprising, especially when combined with songs created with AI with other apps. One of the most popular examples is a children’s song whose lyrics were transformed into a video.
As usual, the first videos created with Jimeng AI contain some errors, especially in the hands and limbs of the characters, and fail to correct one of the big problems of this technology, the corruption of moving faces. But it is still an impressive achievement, considering for example the level at which this technology was last year.
Jimeng AI is currently only available in China, and ByteDance has not announced plans to launch it in Europe and other countries where TikTok is available.
