(MENAFN) China has made great strides in developing AI models, and Kuaishou's Kling is a good example of the rapid progress. Earlier this year, OpenAI unveiled its Sora video production model, which can create one-minute high-definition videos. While Sora is not yet widely available, Kling, a model that creates videos from separate text, is starting to gain traction.
Kuaishou Technology, a well-known Chinese short video platform, has launched Kling, a text-to-video model. Unlike Sora, Kling is available to users through a waiting list. Leveraging similar technology to Sora, Kling can create 1080p HD videos up to two minutes long. The model is based on a Diffusion Transformer architecture that converts text prompts into dynamic images and videos, allowing it to generate realistic, large-scale movements that accurately simulate the physical world.
Kling employs advanced 3D facial and body reconstruction powered by Kuaishou's 3D VAE technology, allowing it to create videos with a variety of aspect ratios. A demo video released by the company showed that Kling can produce videos up to two minutes long at 30 frames per second. In comparison, Sora is currently limited to one minute of videos. Sample videos posted on social media platform X highlight Kling's impressive ability to accurately replicate real-world physics.
It is worth noting that Kling is not Kuaishou's only text-to-video conversion model. In April, the company released Vidu AI, a model that can generate 16-second videos with 1080p resolution, demonstrating China's continued efforts and rapid progress in the field of AI-driven video production.
MENAFN10062024000045015839ID1108313716
Legal Disclaimer:
MENAFN provides the information “as is” without warranty of any kind. We do not accept any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information contained in this article. If you have any complaints or copyright issues related to this article, please contact the provider above.
