ByteDance, the company behind TikTok, has introduced Seedance 2.0, its latest AI model designed for video generation. This advanced technology can create short 15-second clips based on text, images, audio, and video inputs. According to ByteDance, Seedance 2.0 represents a significant improvement in generation quality, particularly in producing complex scenes with multiple subjects while following specific instructions.
Users can enhance their text prompts by providing up to nine images, three video clips, and three audio clips to the model. Seedance 2.0 can generate the clips complete with audio, taking into consideration camera movements, visual effects, and motion dynamics. Additionally, the model can interpret text-based storyboards to create visually engaging content.
This announcement comes as the AI-powered video generation landscape becomes increasingly competitive. Google’s Veo 3 and OpenAI’s Sora 2 have also made advancements in this domain, introducing features like audio support and hyperreal motion capabilities. Runway, an AI startup, has claimed to achieve exceptional accuracy with its latest AI video model.
ByteDance showcased the capabilities of Seedance 2.0 with examples such as choreographed figure skaters performing intricate routines, demonstrating the model’s ability to accurately simulate high-difficulty movements within the constraints of real-world physics. Social media users have already begun exploring the potential of this tool, sharing AI-generated videos featuring famous personalities engaged in cinematic sequences.
Source: The Verge