The answer to VEO and Sora from China. KLING, the latest text-to-video platform from Kuaishou, a Beijing-based company that competes with TikTok, eclipses everything that has gone before. We take an in-depth look at KLING’s features, compare it to other AI video generators and discuss its potential impact on the future of video production.
What is KLING?
KLING is a state-of-the-art AI video generator capable of creating videos up to 120 seconds long in 1080p resolution – at an impressive 30 frames per second. The flexibility of the aspect ratios is an additional plus. But what really sets KLING apart is the ability of its AI model to understand physical relationships and precisely simulate complex movements.
An interesting fact is that eight NVIDIA A100 GPUs are needed to create a one-minute clip on the Sora platform, each costing over USD 10,000. KLING could theoretically require twice as much computing power to generate a two-minute video.
Comparison with other platforms
KLING is in direct competition with other AI video generators such as Sora and Vidu. While Vidu already delivers impressive results, KLING sets new standards in terms of the quality and length of the videos generated. Here are some notable comparisons:
Sora: Although Sora offers more detailed subjects and better lighting conditions, KLING scores with its ability to create longer videos while maintaining narrative coherence.
Vidu: KLING surpasses Vidu with its more advanced AI physics engine that allows for more realistic and fluid movements.
Google’s Veo and others: KLING also beats Veo and other AI video tools such as Pika Labs, RunwayML and StableVideo in terms of image quality and motion display.
Example videos and demonstrations
To demonstrate the capabilities of KLING, here are some sample videos that illustrate the diversity and precision of the platform:
A Chinese man eats noodles with chopsticks: The temporal coherence in this video is remarkable, meaning that the man’s movements and the change in the bowl of noodles appear logical and consistent over time.
A Chinese man sits at a table and eats noodles with chopsticks pic.twitter.com/VnSKgwrtyx
– 青龍聖者 (@bdsqlsz) June 6, 2024
A giant panda plays guitar by the lake: The panda’s gentle movements and realistic interaction with the guitar demonstrate the impressive capabilities of AI.
Panda playing the guitar pic.twitter.com/JLwntfzZea
– 青龍聖者 (@bdsqlsz) June 6, 2024
A man riding in the Gobi desert sunset: a scene of cinematic quality that proves that KLING is not only suitable for simple animations, but also for complex scenarios.
A man riding a horse through the Gobi Desert with a beautiful sunset behind him, movie quality. pic.twitter.com/svdmPIIRMu
– 青龍聖者 (@bdsqlsz) June 6, 2024
Future prospects and availability
KLING is currently available as a public demo in China and can be tested via the Kwaiying app. Although it is not yet accessible worldwide, there are many indications that it is likely to be more widely available by the end of 2024 or early 2025. The developers are focusing on further improving performance and capabilities in order to become an international leader.
The rapid development in AI video generation technology suggests that platforms like KLING could soon play a central role in video production.
