Have you heard of ShengShu-AI? Me neither - until now. The Beijing-based company was only founded last year and specializes in the development of multimodal AI models. Despite its short history, ShengShu-AI has already reached a valuation of 100 million US dollars and attracted well-known investors such as Ant Group and Baidu. A clear sign that something big is emerging here.
Vidu: China's answer to OpenAI's SORA
With Vidu, ShengShu-AI is now presenting a text-to-video india number dataset model that is reminiscent of OpenAI's SORA . According to the company, Vidu can generate videos with a length of up to 16 seconds and a resolution of 1080p. The company uses its own Universal Vision Transformer model (U-ViT) for this.
In a demonstration video, Vidu shows some impressive results that are reminiscent of the release of SORA. In addition to Chinese motifs such as pandas and dragons, there are also many scenes that we already know from SORA. In direct comparison, SORA's results still seem a bit more realistic and detailed, but Vidu is definitely a serious competitor.
Asia's progress in AI development
Vidu is just one of many exciting AI projects to come out of Asia in recent months. Companies like Baidu, ByteDance and Alibaba continue to surprise with impressive developments in the areas of image, audio and 3D generation.
For me, this is a clear sign that Asia may not be as far behind the West in AI research as is often assumed. Perhaps we in the West simply do not get everything that is happening in Asia. I am sure that there are still some surprises in store for us in the future.
Impressive, but not yet usable
As impressive as the progress made by Vidu and SORA is, we should not forget that both models are not yet available to end users. Apart from a few statements from the companies, we still don't know anything about the actual costs, speed and final quality of the generated videos.
One thing is certain, however: the race for the best text-to-video AI continues. I expect we will see a few more new video models in the coming months. Perhaps we are just witnessing the beginning of a development comparable to GPT-1 for video. The future remains exciting!