It feels like we're at a tipping point for AI video generators: Just months after OpenAI's Sora took social media by storm with its text-to-video conversion feature, a new Chinese rival is taking social media by storm.
The new “video generation model,” called Kling AI, was developed by Chinese TikTok rival Kuaishou and is currently only available as a public demo in China via a waiting list, but that hasn't stopped it from spreading quickly, and some impressive clips suggest it's at least as capable as Sora.
Some of the early demo videos (like the one below) can be seen on Kling AI's website, and numerous threads by people like Min Choi (below) on X (formerly Twitter) have compiled impressive early work produced by the tool (with the help of editing apps).
As always, be careful with these early AI-generated clips, as they're only select examples and we don't yet know anything about the hardware or other software used to create them.
For example, later we saw the impressive Air Head The video, which appears to have been created by OpenAI's Sora, required a lot of additional editing in post-production.
This is awesome. While OpenAI Sora sleeps, Chinese AI KLING is breaking the internet. People with access are already generating AI videos and short films. The videos look messed up. 1. “Zootopia Grand Prix” pic.twitter.com/pmCZctsMtTJune 9, 2024
Still, these caveats aside, Kling AI certainly looks like another powerful AI video generator. Early testers have been able to use it to create 1080/30p videos up to two minutes long. The results are surprisingly diverse and promisingly consistent, although there are still some typical AI characteristics, such as smoothing and minor artifacts.
It remains to be seen how long it will take for Kling AI to open up to users outside of China, but OpenAI has suggested that Sora will be available to the public “later this year,” so if Kling AI wants to become the TikTok of AI-generated videos, you probably won't have to wait too long.
The AI video wars heat up
With AI photography tools like Midjourney and Adobe Firefly becoming mainstream, it's clear that video generators will be the next big AI battleground — with huge implications for social media, the film industry and whether we can trust what we see during major election campaigns, for example.
Other examples of AI generators include Google Veo, Microsoft's VASA-1 (which can create realistic talking avatars from a single photo), Runway Gen-2, Pika Labs, etc. Adobe is soon showing how many of these tools can be integrated into Premiere Pro, which will give the field even more impetus.
None of it is perfect yet, and it's not clear how long it will take to create a clip using the likes of Sora or Kling AI, or how much computing power it will require, but the leaps toward photorealism and real-world physics simulation have been so far in the past year that it's clear that we aren't far away from seeing these tools go mainstream.
The fight will also be international, with the US still threatening to ban TikTok, and companies like Kling AI are likely to have to go through some twists and turns before they can scale globally.
