Keling AI recently announced the update of its artificial intelligence video generation tool to version 1.5, introducing a series of impressive new features and improvements. This update significantly enhances the quantity and quality of video generation, further expanding the application boundaries of AI in the creative media field.
Version 1.5 has achieved significant breakthroughs in video generation capabilities, allowing users to generate up to 4 videos at once, and the image-to-video feature now supports generating 1080p HD videos up to 10 seconds long. This upgrade not only increases video output but also enriches and details the video content.
In terms of image quality, Keling AI 1.5 has undergone substantial optimization, now supporting the generation of 1080p HD videos. Whether it's color saturation, detail clarity, or overall visual effects, significant improvements have been made.
Even more exciting is the addition of the "Motion Brush" feature in Keling 1.0's image-to-video, which allows users to specify motion paths for characters, objects, etc. in the image, achieving more precise motion control and more vivid motion representation. This feature provides video creators with more creative space, enabling them to produce more dynamic and engaging video content.
For example, with the moon below, you simply need to select the moon's main body and draw the motion path, and then it will move along the path.
In terms of text comprehension, Keling AI 1.5 has also seen significant improvements, now being able to more accurately understand and execute user text commands, with dynamic quality notably enhanced. Compared to version 1.0, version 1.5 has seen significant improvements in image quality, dynamic performance, and compliance with text commands, with internal evaluations showing an overall effect improvement of 95%.
For example, with the prompt:
3D animation, a fox and a dinosaur dancing under a tree
Version 1.0 produced this, where the fox and the dinosaur start biting each other mid-dance.
While version 1.5 better adheres to the prompt, accurately having the fox and the dinosaur dance under the tree:
In terms of aesthetics and character performance, version 1.5 also demonstrates significant progress. The consistency of characters has been strengthened, maintaining natural and coherent movements even during large-scale actions without strange deformations. Additionally, character expressions are more rich and delicate, showing higher completion and expressiveness in both emotional expression and action details.
For example, with the prompt:
A Chinese girl slowly looking into the camera, with a blurred city nightscape in the background, the protagonist illuminated by artificial light from the front, emphasizing facial contours, with the camera slowly moving to the protagonist's face
The effect generated by version 1.0 was:
The character's face appears oily, and the girl's appearance is not very oriental.
In comparison, version 2.0 offers more delicate images and more aesthetically pleasing compositions:
Consistency of objects has also been improved in the new version, maintaining high consistency and stability in structure and appearance even when objects rotate or move within the frame.
The specific differences between the two versions are as follows:
It is worth noting that the Keling 1.5 model is not available to free users, as its default generation mode is high quality and requires a membership to experience.
Interested users can experience it here: https://top.aibase.com/tool/keling-ai