AGEofLLMs.com
Search

Kling is Now Available Worldwide

Calculating... Comments

Kling AI is a new text-to-video model producing high-quality, realistic videos. When it was first introduced by Kuaishou Technology, a Chinese company known for short-video platforms, it made a big impact in the AI community.

High-Definition Ultra-Realistic Image by Kling
High-Definition Ultra-Realistic Image by Kling

People were impressed by Kling's ability to generate videos up to 2 minutes long in1080p resolution at 30 frames per second. It could accurately simulate real-world physics and produce lifelike motions, something many other AI models struggled with.

The launch of Kling AI was seen as a major development, positioning it as a strong competitor to OpenAI's Sora, which had gained attention earlier. Users who tried Kling praised the quality of the videos it produced from their text prompts. 

But since yesterday, the video generation was open worldwide. Very timely, since I've just received my Chinese sim card to try to gain entry, haha. Nevermind, that sim appears to be broken anyway.))

Kling Worldwide Acccess

Kling AI is now open to everybody around the world and you don't need a Chinese phone number any longer to sign up. Head over to klingai.com and click Sign In (then Sign Up option will show up there). You only need an email, and to confirm it by code they send you. When you first sign in you will see that you have received 64 credits which will expire within 24 hours.

The company is getting paid memebership ready soon. Seen on Chinese version of the website the plans will be quite affordable, at least annual versions, but let's see what they offer their international audience, who knows if those prices were for domestic customers only.

kling-subscriptions-are-coming
Paid Membership Plans Are Coming

When you open video generation section you will find that you can have text or image prompt. Also apparently the end frame is coming soon, so they will be launching keyframes feature similar to what Luma Labs has on their platform. So users will be able to upload starting image and end image, they will drive the interpolation process to create a more precise generation.

kling-generating-video-screenshot
Kling's Video Generation Progress

When you click "Generate 10 credits" video wi'll start processing and you will see a progress bar moving slowly with the message " your creation is brewing enjoy a coffee break while we finalize it " which is kinda hilarious because what I'm doing right now is generating the cup of coffe video! I am trying to make this cup of coffee rotate 360 degrees which I wasn't able to achieve with either Luma Dream Machine or Runway's Gen 3. So for the purpose of scientific curiosity - lol - I'm comparing which platform might be able to do the same task.

You will also notice that a 10 second videos are in the pipeline but they are not currently supported so you can generate a 5 second video. This will cost you 10 credits so you expire will have an opportunity to try out videos like that. You can adjust the settings of between creativity and relevance by dragging the slider. The high performance or high quality mode adjustment is not available right now but what it is is high-performance means faster generation speed and high quality means better visual quality of the outcome. I guess the use for faster lower-quality outputs is that you can test the model's capabilities and how it responds to different prompts and settings. Otherwise of course people will tend to prefer higher resolution video clips.

My First Kling Video Generations

Only a couple of minutes it took, so it seems they've prepared for the influx of users.

Image-to-Video Tests

Coffe Cup Rotating

So my first ever result came out pretty good, and it IS actually rotating the damn cup, can you believe it? 

kling rotating an object
Kling's Coffe Cup Rotation (Lower Resolution Version)

Wasted several credits trying to make the cup spin in Luma yesterday. Nope. Only does that with 2 keyframes but then it adds a second cup handle. Tried in Runway - also no luck spinning it. Kling does it first try. Used the same prompt as in Luma:

Close-up shot on a coffee cup, smooth 360-degree rotation around the cup, showing it from different angles, with subtle lighting changes to highlight textures and details as coffee keeps steaming.

Actually, in Luma preferred wording is 'spin' not rotation but this should kind of be an acceptable close synonym? Kling gets it just fine...

Anyway, will carry on testing Kling and update this later.

Cows Circle Dancing

second-image-to-video-test-cows
Second Image-to-Video Try

This one has been my obsession to make in Luna but it never happened. Well, I must say, this comes very close to it! Nothing remotely similar was produced in Dream Machine in like a dozen attempts.

kling making cows dance
Whimsical, surreal view of a ritualistic circle dancigng by cows.

Notice how there's no disfigurement, accurate animation of completely unnatural cows body movements.

Tiny Kitten in a Palm Moews

This one is high quality Midjourney generation of mine. The prompt for Kling was:

cute tiny cat meows softly while moving lightly in slow motion

Which it has followed for the most part. The cat stayed silent, no meow.

kling animating tiny kitten
Tiny Kitten in a Palm

Text-to-Video Tests

Kling's Cornfield Text-to-Video

This was my prompt:

A vast cornfield bathed in golden hour light, highlighting the rich colors of the corn stalks. Focus on the scarecrow standing prominently in the middle as it starts moving, suddenly coming alive. The sky above transitions from warm orange to deep purple as the sun is setting. Eerie, surreal.

This one turned out kind of underwhelming. The scarecrow never moved a 'muscle'. Although the general view of the cornfield and sunset came out nice.

Sportscar Driving at Night Text-to-Video

Tracking shot: A sleek sports car driving on a reflective wet road at night. Neon lights from nearby signs create vibrant reflections on the surface as the car speeds past, highlighting its movements and the glowing colors.

Well, ok, but not great. I've had a more breathtaking output for this prompt in Runway. And Haiper was pretty good with it too to be honest. I've even reformulated my prompt a bit, made camera movement automatic (up to AI to decide) and got an even worse view with darker, gloomier road, even fewer neons.

So far I'm finding image-based generations more reliable with Kling as well as with other models that have this capability. When they let us purchase subscriptions where both keyframes will be available - this should be the most exciting time.

For now, here's :

American Gothic Animation

by Kling. From picture, of course.

Related Posts

Visitor Comments

Please prove you are human by selecting the tree.