China's large video generation model now accessible worldwide
The China-developed large video generation model is now accessible for users worldwide.
TroibNews 
Vidu, a significant video generation model developed by Chinese AI firm ShengShu Technology and Tsinghua University, is now available for global use. This model supports both text-to-video and image-to-video generation capabilities.
Vidu can produce 4-second clips within 30 seconds and generate videos up to 32 seconds in length in a single instance.
"Vidu can simulate the real physical world, creating detailed scenes that adhere to physical laws, such as natural lighting and shadow effects, as well as intricate facial expressions. Additionally, it can generate surrealistic content with depth and complexity," said Zhu Jun, deputy director of the Tsinghua Institute for Artificial Intelligence.
Zhu also mentioned that Vidu can tailor scenes to various genres like sci-fi, romance, and animation, capturing the essence of each style while also providing high-quality cinematic effects, such as smoke and lens flares.
The AI model is versatile in managing different shot types, from long shots to close-ups and medium shots, and can effortlessly produce effects like long takes, focus pulls, and smooth scene transitions.
Users have the option to upload portraits or custom character images and utilize text descriptions to direct these characters to perform any action in any scene, streamlining the video production process and fostering creative freedom.
According to the company, Vidu's core architecture was conceptualized as early as 2022. The AI model was officially introduced at the 2024 Zhongguancun Forum in Beijing this past April, about two months after OpenAI revealed its Sora video model. Despite its introduction, Vidu has maintained a low profile since the forum.
In the interim, similar tools, including Kuaishou's generative video model Kling and the large language model family ChatGLM, have been made available to users.
(With input from Xinhua)
Thomas Evans contributed to this report for TROIB News
Vidu can produce 4-second clips within 30 seconds and generate videos up to 32 seconds in length in a single instance.
"Vidu can simulate the real physical world, creating detailed scenes that adhere to physical laws, such as natural lighting and shadow effects, as well as intricate facial expressions. Additionally, it can generate surrealistic content with depth and complexity," said Zhu Jun, deputy director of the Tsinghua Institute for Artificial Intelligence.
Zhu also mentioned that Vidu can tailor scenes to various genres like sci-fi, romance, and animation, capturing the essence of each style while also providing high-quality cinematic effects, such as smoke and lens flares.
The AI model is versatile in managing different shot types, from long shots to close-ups and medium shots, and can effortlessly produce effects like long takes, focus pulls, and smooth scene transitions.
Users have the option to upload portraits or custom character images and utilize text descriptions to direct these characters to perform any action in any scene, streamlining the video production process and fostering creative freedom.
According to the company, Vidu's core architecture was conceptualized as early as 2022. The AI model was officially introduced at the 2024 Zhongguancun Forum in Beijing this past April, about two months after OpenAI revealed its Sora video model. Despite its introduction, Vidu has maintained a low profile since the forum.
In the interim, similar tools, including Kuaishou's generative video model Kling and the large language model family ChatGLM, have been made available to users.
(With input from Xinhua)
Thomas Evans contributed to this report for TROIB News
