Vidu 1.0

Advanced Text to Video

Vidu 1.0's text-to-video feature allows users to generate high-definition (1080p) videos lasting up to 16 seconds from simple text prompts.

Built on a self-developed visual transformation model architecture known as the Universal Vision Transformer (U-ViT), Vidu 1.0 integrates two powerful text-to-video AI models: the Diffusion and the Transformer, allowing it to simulate real-world physics, intricate facial expressions, and dynamic camera movements, resulting in videos that are not only aesthetically appealing but also contextually rich.

Powerful Image to Video

The image-to-video feature of Vidu 1.0 offers an innovative way to animate still images into dynamic video content in just seconds. Users can upload an image and leverage Vidu's advanced algorithms to generate animations that maintain the original context while infusing it with creativity.

Input Text	Input Image	Output Video
A blonde woman with blue eyes walking along the beach.

Reference to Video 

Vidu 1.0's reference-to-video feature allows users to create character consistent videos. This capability is essential for creators who require coherence in subjects, settings, and visual styles.

By allowing users to upload reference images, Vidu AI ensures that characters and objects maintain their appearance throughout various scenes.

Users can also upload reference images of any environment, and utilize descriptive keywords to introduce new elements into the scenes, whether it's a character, animal, or object.

Moreover, Vidu AI's tool goes beyond simple character consistency. It enables creators to splice together various subjects and environments effortlessly.

Input Text	Input Image	Output Video
A mysterious blue creature with long ears crawls through the forest, surrounded by the quiet ambiance of the night. The camera moves backward, capturing a close-up shot.

Various Video Templates

Vidu AI offers a wide variety of engaging video templates to choose from, allowing users to create dynamic and creative content with ease. These templates include unique AI-powered features like AI hugging and AI kissing, which bring a touch of playfulness and innovation to video generation.

Input Text	Output Video

Vidu AI's Team, Technology And Impact

Vidu 1.0 is driven by the universal vision transformer (U-ViT) model, developed by chief scientist Zhu Jun and his team at Shengshu. Introduced in a 2022 research paper, U-ViT combines transformer and diffusion algorithms, creating a robust architecture for generating diverse video outputs.

Since its launch, Vidu AI has gained attention in the film industry. Notably, Chinese director Li Ning is reportedly using Vidu AI and other generative AI tools to produce China's first fully AI-generated movie, set to release later this year. The platform's capability to maintain visual consistency across scenes is crucial for this innovative project, showcasing the potential of AI in transforming future filmmaking.