Developments in the field of artificial intelligence are expanding the boundaries of the technology world day by day. Among these innovations, a model that challenges our imagination in video production came from China. Kuaishou, TikTok's rival in China, has developed KLING, a new open-access model that is similar to OpenAI's Sora model, which is not yet widely available, but performs better in many respects. This model, named KLING, has the ability to create almost realistic videos from texts.
A new breath in artificial intelligence-supported video production with KLING
With a simple depiction of a Chinese man sitting at a table eating noodles with chopsticks, the KLING model was able to produce an almost lifelike video that compares favorably to Will Smith's video of demonic-looking noodles created by Modelscope Text2Video last year.
KLING can produce 2-minute long videos at 1080p quality and 30 frames per second (fps) with a single command. It can accurately simulate real-world physical properties. Using Diffusion Transformer architecture, KLING transforms rich text descriptions into live scenes. Supporting various aspect ratios with dedicated 3D VAE support and variable resolution training, the model offers the opportunity to capture full expression and limb movement from a single full-body photograph, with advanced 3D face and body reconstruction technology.
It is clear that China is among the world leaders in developing artificial intelligence models. Now open access, KLING offers just one example of how far the country has come in this field.
OpenAI has announced that it plans to launch Sora by the end of the year, but it is debated whether this effort will be sufficient in the face of China's text-to-video conversion models. OpenAI's biggest advantage in this race may be the possibility of China not making its model available worldwide.
Interestingly, the KLING is not the first video production model to come out of China. Released in April, Vidu AI was the first to be considered the Chinese version of Sora, capable of creating 16-second long videos with 1080p resolution.
These technological advances are expanding the boundaries of AI-powered creativity and offering exciting clues about how content production will shape the future.