Chinese startup Senshu AI, a self-proclaimed “Sora rival,” has launched its video-generating AI globally. The model is capable of making 4-second clips in 30 seconds, in line with the present trend, and competitors are launching official services one after one other ahead of Sora.
The South China Morning Post recently reported that Senshu has launched a world service for its video-generating AI, Vidu, which supports each Chinese and English.
This model produces two clips, 4 and eight seconds long. Currently Through the official website You should utilize it. You’ll be able to join together with your Google ID, email, or phone number.
Senshu was the primary of the Chinese startups to unveil its product within the recent fierce chase after Sora. It first unveiled Bidu on the Zhongguancun Forum in April, two months after Sora’s debut.
On the time, Senshu explained that it was based on a visible transformation model architecture called ‘Universal Vision Transformer (U-ViT)’ that integrated the diffusion model and transformer architecture, and that this was very just like Sora.
This time, we have added a ‘character-video’ feature to the text-to-video feature. This helps you to upload a picture of an individual or an animated character and generate a video with an easy text prompt.
“In the longer term, we hope to create videos just like how movies are made, by having users upload multiple characters and describe scenes,” Senshu said. “Our goal is to integrate AI tools with existing fields.”
The corporate has already received tens of hundreds of thousands of dollars in investment from Baidu, Alibaba and others.
Meanwhile, at the top of last month, Kuaishou’s ‘Cling’, which boasts the best quality in China, expanded right into a paid version, and a couple of days later, startup Jifu AI also released its ‘Ying’ model, which generates a 6-second video in 30 seconds, as a web site and app.
On the first of this month, Runway announced that it is going to be releasing the ‘Gen-3 Alpha Turbo’, which creates a 10-second video in 11 seconds, to a small variety of test subjects and might be released to paid users as early as this week.
Reporter Im Dae-jun ydj@aitimes.com
