Moments Lab has launched MXT-2, a new version of its multimodal AI indexing sysyem designed to help video producers navigate and extract content from large video libraries.
Moments Lab has launched MXT-2, a new version of its multimodal AI indexing technology designed to help video producers navigate and extract content from large video libraries. The updated technology focuses on streamlining the process of content creation, repurposing and monetisation by making it easier to identify key moments within videos.
MXT-2's dataset has grown significantly, now containing over 1.5 billion images, more than triple the size of its predecessor. This expanded dataset allows the AI to generate more accurate and detailed descriptions of video content, giving users more refined insights into their video assets.
One notable feature is the introduction of Custom Moments, which are time-coded descriptions that highlight specific video segments tailored to user needs. This tool is particularly useful for identifying standout scenes in unscripted TV shows, such as the most engaging moments for creating short clips, trailers or compilation videos.
Another new addition is the Custom Insights feature, used to automatically generate written content of various lengths ranign from titles and teaser descriptions to full-length articles. Custom Insights can also classify videos by theme or editorial topic, further assisting in content organisation.
Frederic Petitpont, the CTO and co-founder of Moments Lab, emphasised the capability of MXT-2 in video sequencing - arranging clips into a specific order that conveys the user's narrative or idea - noting its favourable performance in this area against Google Research's Vid2Seq. Philippe Petitpont, CEO and co-founder, commented on the company’s goal of simplifying video content management, stating that MXT-2 enhances the searchability and editorial insights of large video libraries, offering users greater opportunities for content reuse and repurposing.
MXT-2 is now available on the Moments Lab platform. Alternatively, users can integrate it into their existing workflows and tools through an API.
The technology will be demonstrated at the 2025 NAB Show, taking place 6-9 April in Las Vegas, at the Moments Lab booth (SL12213).
Furthermore, Moments Lab's AI software engineer Khalil Guetari will present at the BEIT Conference during the NAB Show. His session, titled 'Vision and Language Models for Enhanced Archive Video Management' will take place on 7 April from 10.40 to 11am in Room W222-W223. www.momentslab.com
