Chinese firm Kuaishou released text-to-video model "Kling," could be Sora's biggest competitor

Chinese short-video platform Kuaishou has introduced “Kling,” a text-to-video generation model. Kling joins a growing field of similar models that can create videos from textual descriptions, like OpenAI’s Sora and Google’s newly unveiled Veo.

Kling can generate videos in 1080p resolution at 30 frames per second, with a maximum length of two minutes. The model can design portrayals of the physical world, including accurate movements and object interactions.

Kling uses 3D face and body reconstruction to create lifelike characters with a full range of expressions and fluid limb movements. Here is an example of Google’s Veo.

Currently, Kling is available for open access, which means users can experiment with the model’s capabilities. This is in contrast to the limitations on global access for OpenAI’s upcoming “Sora” model.

Kling’s development is a clear example of China’s advancements in AI. But overall, it’s astonishing how far AI has come within less than 2 years. Here is a reference video:

The release of Kling, coupled with the upcoming release of Sora, suggests continued progress in text-to-video generation technology. While the long-term impact and accessibility of these models remain to be seen, their development signifies a dynamic landscape within the field of AI video generation.