China’s Shengshu Technology and Tsinghua University have revealed “Vidu” at the 2024 Zhongguancun Forum. Positioned as a direct competitor to OpenAI Sora, Vidu’s text-to-video model generates 16-second clips at 1080p resolution in seconds.
Vidu operates on a Universal Vision Transformer (U-ViT) architecture, which enables it to simulate complex real-world scenes with multiple camera views. This allows Vidu to transition seamlessly between long shots, close-ups, and medium shots within a single clip, showcasing its dynamic shot creation capabilities.
🚨 China just released SORA’s rival “Vidu”
This is China’s first long duration, high consistency, and high dynamics video modelIt can create videos upto 16s with 1080P in single click.
It excels at simulating the real physical world and also showcases a vivid imagination,… pic.twitter.com/6ThjAxrQs2
— Sambhav Gupta (@sambhavgupta6) April 27, 2024
By comparison, OpenAI’s Sora can generate videos up to 60 seconds, although Vidu’s quick generation is its own advantage.
The model’s advanced rendering capabilities give its videos a high degree of realism, incorporating intricate lighting, shadows, and facial expressions. Additionally, Vidu has demonstrated its ability to create surreal content, adding depth and imagination to its videos.
Shengshu Technology’s announcement emphasized Vidu’s ability to produce complex videos with realistic lighting and transitions. In its demo, Vidu recreated scenes similar to those showcased by OpenAI Sora.
Although Vidu’s output is yet to reach Sora’s level of visual fidelity, it boasts remarkable temporal consistency, allowing for smoother and more coherent clips.
The Vidu model exemplifies China’s rapid strides in Artificial Intelligence research, marking a milestone for the country’s tech industry. While Sora may currently hold a slight edge in video quality, Vidu’s entry into the market is a strong statement of China’s AI capabilities and signals more competition to come.
Vidu’s progress could also accelerate enhancements in Video Generative AI, pushing the boundaries of what can be achieved in seconds. This competition between Vidu and Sora reflects the evolving dynamics of AI video technology, promising ongoing improvements and innovations that will benefit consumers and industries alike.
Tech enthusiasts over the internet also think it’s good to have competition. Here’s what one person said,
Comment
byu/bemmu from discussion
insingularity
The revealing of Vidu indicates a significant advance for China’s AI industry. The model’s versatility, along with its innovative U-ViT architecture, highlights the growing capabilities of Chinese tech firms.
While Sora remains a strong competitor, Vidu’s entry into the market shows that the field of AI-powered video generation is expanding, presenting new opportunities for creativity and technical progress.
Found this interesting? Visit allaboutai.com for more exciting AI News like this.
