Google Unveils Lumiere: A Revolutionary AI Model for Real-time Video Creation
Summary:
Google researchers have unveiled Lumiere, a new diffusion model that can instantly create and modify realistic AI videos from text or static images. This breakthrough model is seen as a game-changer in AI-video technology, distinguishing itself with a single-pass approach to video creation from existing AI models. Built on a dataset of 30 million videos and captions, Lumiere can currently produce 80 frames at a speed of 16 frames per second.
Google's team of scientists has unveiled their project Lumiere, a novel diffusion model that can create realistic AI videos from text or still images, including competing on-the-go modifications. Lumiere specializes in depicting "authentic, varied, and consistent motion" using its "Space-Time U-Net architecture", allowing it to instantly generate an entire video through a single pass of the model. Google has achieved an extraordinary advancement in AI-video technology with this diffusion model, Lumiere, setting 2024 up to be a pivotal year for AI video.
In the scientific report, the researchers elaborated that: “Our model employs spatial and crucially, temporal down- and up-sampling and draws on a pre-trained text-to-image diffusion model to instantly produce a full-frame-rate, low-resolution video, as it processes in multiple space-time scale”. This suggests that users can enter a textual description of their desired video content or upload a static image together with a command to create a live video. Lumiere has been likened to ChatGPT, but with the added abilities of video creation, stylizing, editing, and animation from text and images, the paper suggests. The Google team's unique method of using a single-pass for handling time-related data in video creation distinguishes Lumiere from existing AI video generators like Pika and Runway. Related: AI deepfakes deceive citizens and politicians prior to US 2024 elections - 'I believed it was genuine' A sample demonstration of Lumiere's capabilities was shared on X by Hila Chefer, a student researcher who collaborated on the model development with Google.
Lumiere was developed based on a dataset of 30 million videos along with their descriptive captions; however, the video data source, a contentious issue in the AI and copyright law realm, has not been referenced yet. At present, the model is capable of producing 80 frames at a speed of 16 frames per second. Users on X have lauded Google's achievement as "outstanding", "cutting-edge", and they're anticipating a significant surge in video creation in the upcoming year.
Magazine: Selections of Crypto+AI tokens, Decades required for AGI, Expansion of Galaxy AI to 100M phones: AI Eye.
Published At
1/25/2024 5:00:00 PM
Disclaimer: Algoine does not endorse any content or product on this page. Readers should conduct their own research before taking any actions related to the asset, company, or any information in this article and assume full responsibility for their decisions. This article should not be considered as investment advice. Our news is prepared with AI support.
Do you suspect this content may be misleading, incomplete, or inappropriate in any way, requiring modification or removal?
We appreciate your report.