A New Chinese AI Video Generator "Kling" Is A Threat To OpenAI's Sora

Kling can generate up to 120-second videos with 30 frames per second at 1080P resolution and free aspect ratio.

In one of my previous articles, I featured a Chinese AI video generator called Vidu. I called it a real Sora competitor because of how impressive the sample videos were. Today, another AI video generator tool called Kling was unveiled, and it looks even better than Vidu.

What is Kling?

Kling is a new AI video generator from Kuaishou (“quick hand”), a company from Beijing that competes with TikTok.

Kling can generate up to 120-second videos with 30 frames per second at 1080P resolution and free aspect ratio. According to its creators, their AI model can understand physics better and model complex motion accurately.

Here’s a fun fact: Sora requires eight NVIDIA A100 graphics processing units (GPUs) running for over three hours to produce a one-minute clip. One NVIDIA A100 costs over 10,000 USD. So Kling would probably require double of that compute power to produce a 2-minute video result.

Take a look at this example video:

Video example from Kling

Prompt: A Chinese man sitting at a table, eating noodles with chopsticks

You can see how good the temporal coherence on this example video.

Temporal coherence in AI videos refers to the ability of a video generation model to create a sequence of frames that are consistent and logically connected in terms of time.

This means that the model should be able to maintain a consistent narrative, maintain the same scene or setting, and ensure that the actions and movements of objects within the scene are coherent and plausible over time.

You can explore the website and be amazed by the examples. You can also check the example GIFs I attached below. Kling is currently open for testing on Kuaishou’s video clip app Kmovie.

How does it compare with Sora?

Actions that affect the state of the world are some of the hardest simulations that an AI video generator faces. For example, a painter can leave new strokes along a canvas that persist over time, or a man can eat a burger and leave bite marks.

Both Sora and Kling can do that.

So why not put them side by side? Here’s an example video of a person eating a hamburger: