Kling AI made waves when it was released last month, offering high-quality generative video content on par with OpenAI's yet-to-be-released Sora. It was initially only available in China and required a Chinese phone number, but has now expanded worldwide.
Developed by Chinese tech giant Kuaishou Technology, the Kling model can generate up to two minutes of video in HD resolution, and uses advanced 3D reconstruction technology to create more natural movements of humans and animals.
In an article about X, the Kling team announced the international version, which will require 10 credits for each generation and will give 66 credits per day, with subscription levels coming soon.
What is Kling and why is it important?
🎉 The moment we've all been waiting for has arrived! 🎊 Introducing the official global launch of the international version 1.0 of Kling AI! 🌍📧 You can log in with any email address, no mobile number required! 👉 Direct link: https://t.co/68WvKSDuBg 🔥 Log in every day and get 66 free credits… pic.twitter.com/TgFZIwInPgJuly 24, 2024
Artificial intelligence video generation is a rapidly changing and growing field: Runway brought increased realism and 10-second clip generation with Gen-3 earlier this month, Luma Labs introduced the impressive Dream Machine last month, and Haiper has also been improved.
OpenAI is currently only allowing access to Sora to a select few creators, so Kling may be the champion.
These new model updates were initially prompted by OpenAI's launch of Sora earlier this year, which still stands out as the best content generator around, with a slew of new short films recently released on YouTube that showcase its full potential, with Kling also contributing to the update flurry.
Kling AI takes a slightly different approach to AI, leveraging generative 3D in the creation process and offering Sora-level scene changes, clip lengths, and video resolution. OpenAI is currently only allowing access to Sora to a select few creators, so Kling might be the champion.
The ability to generate clips up to two minutes long is also important, as Sora takes up to about one minute per generation, with each generation taking upwards of 15 minutes, according to OpenAI.
However, OpenAI can take some solace in the fact that the global release currently only allows five seconds for the first clip to be created, a long time per clip, so unless Kling upgrades soon, Sora still has a good chance once it hits the market in the coming months.
How do I access Kling?
There are two ways to access Kling. The most powerful and feature-rich version is only available in China and requires a Chinese phone number. There are services that will allow you to get a phone number, but they are unreliable and not recommended.
The other option is to accept the limitations of the global version, which will catch up with the Chinese version release pretty quickly. In this case, just go to KlingAI.com, sign in with your email address, and start creating videos.
It offers text-to-video and text-to-image conversion capabilities in an easy-to-use interface, and unlike Runway, Haiper, and Luma Labs, it lets you enter prompts of up to 2,000 characters, allowing you to create very detailed descriptions. It seems to work better with longer, more thoughtful prompts.
One thing to note is that it's slow. It took me less time to write this entire article, edit it, enter it into our content management system, and publish it than it took to create the first 5-second video. This may be because Dream Machine was similarly slow on initial release, so everyone rushed to try it out.
Final thoughts
Kling's global release is a good thing for the AI video market, as competition is likely pushing other companies to update and release new models more quickly.
Ultimately, all AI video tools will settle on a similar quality level as they aim to create an “open world” AI platform, where the underlying models understand the real world.
It all comes down to additional services, production speed, and most importantly, price.
More articles from Tom's Guide
Back to MacBook Air
