Recently, the first purely self-developed original video large model in China, Vidu, jointly developed by Shengshu Technology and Tsinghua University, has introduced the "Subject Reference" feature. This functionality enables consistent generation of any subject, making video creation more stable and controllable. Currently, this feature is available to users for free.

The "Subject Reference" feature addresses the issue of character consistency in AI-generated videos. By uploading a single image of any subject, Vidu can lock in the subject's appearance and, through descriptive keywords, switch scenes at will, outputting videos with consistent subjects.

QQ20240911-143728.png

This feature is not limited to a single object but is designed for "any subject," whether it be a person, animal, product, or even an anime character or a fictional entity, ensuring consistency and controllability in video generation.

Tang Jiayu, Chairman and CEO of Shengshu Technology, stated that the "Subject Reference" feature represents the beginning of AI's complete narrative capability, and AI video creation will also move towards a more efficient and flexible stage.

Additionally, Shi Yuxiang, a director from China Central Television and an AIGC artist, created an animated short film titled "A Summer Gift," and expressed that the "Subject Reference" feature breaks free from the constraints of static images, resulting in more impactful and free-flowing visuals, significantly enhancing the continuity of creation.

The launch of this feature marks a new era for AI video creation, with future explorations focusing on precise control of multi-subject interactions, consistent styles, and stable switching between varied scenes, to meet higher narrative demands.