"After studying digital film production for 7 years, I now feel that 90% of it was in vain." On social media platform X, “el.cine,” a creator in the AI film and creative content field, declared they were completely shocked after trying ByteDance's new generation AI video generation model Seedance 2.0. ByteDance's latest video generation model, Seedance 2.0, recently began internal testing on products such as Dreamina, Doubao, and Little Skylark, attracting widespread attention both in China and abroad.
It is understood that videos generated by this model can automatically achieve audio-video synchronization and complete dialogues, among other functions. Some users have used it to create short drama clips featuring multi-character dialogues, storyboards, and cinematic camera movements. There are already quite a few demo videos online, including an impressive Dragon Ball duel in the style of Shaw Brothers films.
● “Black Myth: Wukong” Producer: Glad It Comes From China
Jing Feng, CEO of Game Science and producer of "Black Myth: Wukong", after trying Seedance 2.0, called it "currently the world’s most powerful video generation model," stating that its multi-modal comprehension and integration capabilities represent a leap forward, lowering the cost of video creation to nearly the marginal cost of computing power, and predicting it will popularize video creation among the masses. "AI's ability to understand and integrate multi-modal information (text, image, video, audio) has made a giant leap—it's astonishing," Feng said. "I'm glad that at least today's Seedance 2.0 comes from China." However, he also pointed out the challenge of potential fake video proliferation.
In response to user concerns that AI-generated real-person videos could violate privacy, platform operators indicated that currently, uploading real human images or videos as subject references is not supported, and the product will be relaunched with improved features after adjustments.
It is understood that, at present, when using Seedance 2.0 on the Dreamina web platform or Little Skylark, users are prompted that real human face references are not supported; in the Dreamina app and Doubao app, users must record their own likeness and voice for live identity verification before creating a digital avatar to appear in AI-generated videos.
Professor Shalei from the Artificial Intelligence Research Institute at Beihang University commented that Seedance 2.0 is still in a small-scale internal testing phase, with ByteDance restricting some model functions—for example, only allowing generation of real-person videos following live authentication and not supporting user-uploaded photos or footage of real people as subjects—to prevent misuse of AI technology. "This kind of attempt is the responsibility baseline that enterprises must adhere to while driving forward with technology at speed."
He stated that AI progress will never stop due to controversy, but how to achieve a balance between technological innovation and data compliance, as well as copyright protection, so that the dividends from breakthroughs in technology truly benefit the industry and users, is a question that not only ByteDance or a single company must continually explore, but the entire global AI industry must also address.
Seedance2.0 is an AI video generation model launched by ByteDance that can create movie-level videos from text or images. On February 7, 2026, it began small-scale internal testing and was launched on the Dreamina platform, where members can directly use it.