AI Card Bao AICARDBAO.COM

Industry News

Kaling releases Kling Image 4.0: Native 8K output – a single image is capable of displaying up to 2,000 characters of text.

This time, KeLing AI has released not a video model, but an image model.

On November 17th, Kelying AI launched the new generation of image creation model, Kling Image 4.0, offering its APIs to corporate customers and developers. The most prominent feature of this version is its support for native 8K image generation – this represents Kelying AI’s first time supporting such a high resolution.

In terms of its architecture, it integrates both understanding and generation into a unified system. At the practical application level, some numbers can better illustrate this: text input can contain up to 15,000 tokens, and a single image can display 2,000 characters; in addition, it comes equipped with more than 70 commercially available fonts.

What does the concept of “2,000 words” mean?

It’s roughly equivalent to the amount of text that would appear on a long information graphic or a menu. For a long time, the weakness of image processing models has been their handling of text. Handling a few words is still possible, but when the number of characters increases, errors, garbled text, and overlapping strokes are likely to occur. By specifically mentioning the maximum amount of text that can be properly formatted and the use of commercial fonts, it’s clear that the target audience for these recommendations includes e-commerce product detail pages, marketing materials, posters, and other scenarios where a large amount of text is frequently displayed.

Another useful feature is the ability to create multiple images in a consistent style. Up to 10 images with the same design can be generated at once, and it is also possible to produce very long-format images. For those who work on creating series of posters or serialized illustrations, the biggest challenge is ensuring that each image maintains the same aesthetic style. Re-designing each individual image separately takes a lot of time; however, generating an entire set of images at once eliminates this need for repeated editing.

For developers, the upper limit of 15,000 tokens for input also has practical benefits: complete design specifications and entire chunks of text can be inserted at once into the prompt system, eliminating the need to break them down into multiple rounds of back-and-forth communication.

The model has now been fully launched. However, this time it is specifically designed for enterprise customers and developers, who can access it through APIs. Individual users will be able to use it directly only if it becomes available through certain designated channels in the future.

Related