Key features of Vidu include its ability to simulate the real physical world and generate scenes with complex details that adhere to real physical laws. This includes creating reasonable light and shadow effects and delicate facial expressions. Unlike some other models, Vidu can generate complex dynamic shots, not just fixed ones.
The core technology of U-ViT was proposed by Vidu’s research team in September 2022, predating the model architecture of Sora’s DiT - Diversity in Transformation. This makes Vidu a significant advancement in the field of AI and showcases China’s rapid development in this critical emerging field.