Wedoany.com Report on Feb 13th, ByteDance recently unveiled its next-generation video generation artificial intelligence model, Seedance 2.0. This model possesses multimodal input capabilities, allowing it to generate video clips up to 15 seconds long, complete with synchronized audio tracks, based on various prompts such as text, images, audio, and video. This move signifies a further intensification of competition among tech giants in the AI video generation race.

According to ByteDance's official blog, Seedance 2.0 achieves a significant leap in video generation quality, particularly excelling in handling complex dynamic scenes involving multiple subjects, and can respond to user instructions more accurately. Users can upload up to nine images, three videos, and three audio clips as input to refine text prompts. The model holistically considers camera movement, visual style, and action choreography during the generation process and can construct content based on storyboards derived from text descriptions.

Over the past year, AI video generation technology has evolved rapidly. Google introduced Veo 3 with audio synthesis capabilities, OpenAI released Sora 2 and its companion application, and Runway also launched a new version of its model focused on high instruction adherence. Against this backdrop, the entry of Seedance 2.0 further enriches the technological pathways and options available for multimodal content generation tools.

In one demonstration by ByteDance, two figure skaters synchronously performed a series of difficult movements including take-off, aerial rotation, and landing, with trajectories conforming to real-world physical laws. The company stated that the model can "consistently execute complex motion instructions while strictly adhering to the laws of physics in the real world."

Early user feedback on Seedance 2.0 has already appeared on social media. In a widely circulated generated video, actors Brad Pitt and Tom Cruise were placed in a movie-style fight scene. Rhett Reese, co-writer of "Deadpool," commented upon sharing the clip, "I hate to say this. We might be done for."

Other demos show the model's ability to generate various visual forms, including anime, cartoons, sci-fi movies, and even styles imitating digital creators. The copyright protection mechanisms for Seedance 2.0 are currently unclear; searches on platform X reveal that some user-generated clips contain characters from well-known IPs such as "Dragon Ball Z," "Family Guy," and "Pokémon."

Currently, Seedance 2.0 is only available to the public through ByteDance's Dreamina AI creation platform and its AI assistant Doubao. Whether it will be integrated with the short-video platform TikTok has not been disclosed by the company.