Artificial intelligence (AI) researchers from the Google Research division of the American company Google have created a text-based video content creator called Lumiere. The team published a progress report on: platform scientific publications arXive.
According to the statement, the new neural network will be able to create ready-made videos in high resolution in accordance with short queries such as “two raccoons reading a book.”
Google describes the technology behind the new tool as the “revolutionary U-Net space-time architecture.” It is designed to create video in a single session of the artificial intelligence model.
The tool can either create videos from scratch or edit ready-made clips upon request. Lumiere can also animate a static image.
Google hasn’t yet said whether or how soon they plan to make Lumiere a public service. This may be due to legal consequences if Lumiere creates videos that violate copyright laws.
Previously plexus Sbera managed to surpass ChatGPT in the quality of English responses.