Google's unveils it AI video generator 'Lumiere', saying good bye to exhaustive video editing
Technology
AI video maker Lumiere will create video from text and image
(Web Desk) - Google has introduced Lumiere, its new AI video maker, joining the race in the AI field.
Lumiere utilizes a novel diffusion model called Space-Time-U-Net (STUNet) to understand the spatial and temporal aspects of a video, allowing it to create videos in a single process instead of piecing together individual frames.
The system starts by generating a base frame from a prompt and then uses STUNet to predict the movement of objects within that frame, producing a sequence of frames that smoothly transition between each other.
In addition to text-to-video generation, Google's Gemini large language model will eventually bring image generation to Bard. This feature enables users to create videos with specific styles, cinemographs that animate only a part of a video, and in-painting to modify the colour or pattern of a specific area in the video.
While Lumiere is not yet available for testing, it showcases Google's ability to develop an AI video platform that rivals, and perhaps surpasses, existing AI video generators like Runway and Pika.
However, Google acknowledges the potential for misuse, such as creating fake or harmful content with their technology.
The company emphasizes the importance of developing tools to detect biases and malicious uses to ensure safe and fair utilization, although specific methods for achieving this were not detailed in the paper.
