正文 · AI 翻译
IT Home reported on November 17 that Kuaishou's Kling AI announced late tonight (the 17th) that its new generation image creation model, Kling Image 4.0, has officially launched. API access is also available to enterprise customers and developers.
The new model uses a “unified understanding and generation” architecture, exploring the integration of visual understanding and image generation into a unified model system to further connect the process from understanding creative intent to generating visual content. Kuaishou said that, alongside further improvements in model capabilities, image creation now has more possibilities in realism, text rendering and content organization. Kling Image 4.0 brings three major upgrades in these areas:
Realism and native 8K: By incorporating physical principles and visual perception into the generation process, it makes lighting, skin texture and material textures appear natural, close to images straight from a camera. It supports native 8K image generation for the first time, showing richer detail and providing more room for high-definition display and subsequent cropping.
Powerful text rendering: It supports generating 2,000 characters of text in a single image and more accurately understands complex instructions and layout requirements, making small text clear and decorative lettering stand out.
Coherent image sets and long-form images: It plans content according to creative intent, generates up to 10 related images from a single description, and supports long-form images for e-commerce product detail pages and web design, making the creation of complete sets and full-page designs more efficient.
IT Home learned from the official introduction that Kling Image 4.0 supports text inputs of up to 15,000 tokens, enabling it to understand more complete and complex creative requirements and better handle text placement, image layout and subject presentation.
The model includes more than 70 preset fonts for commercial use and supports uploading reference images to guide the shape and visual style of text, bringing generated results closer to the intended design. For languages, it supports Simplified Chinese, Traditional Chinese, English, Japanese, Korean, Spanish and Portuguese, allowing ideas in different languages to be presented more effectively in images.
Kling Image 4.0 can generate up to 10 images from a single description, allowing the entire set to share the same creative requirements while maintaining a consistent overall style and giving each image a distinct focus.
The model's native agentic capabilities can also help plan the entire image set: it understands creative intent and automatically arranges the content and focus of each image, making it easier to develop everything from storyboards to product displays.
In addition, Kling Image 4.0 includes more than 110 preset visual tone styles for switching between different visual effects. Users can also upload images they like as references, extract their color and lighting styles, and create a custom visual tone.