17/12/2025
Was invited for a little pre-Christmas creative and technological experimentation down at the labs last week - exploring their fascinating âAI lensâ video production systemâŚ
Itâs very much an early stage âalphaâ model but the concept around it is absolutely fascinating.
The premise is that this virtual production space can film actors, dancers, puppets, props - anything you want - and simultaneously generate AI footage (30 AI images a second) onto the screen as you film.
All the AI visual input data is controlled, and all energy locally managed. Your performers trigger the visual input and AI develops the video content.
Itâs a refreshing new approach to how you are able to control AI image data input and energy usage. It also means you can use real live human input with live AI integration.
This fascinating day was facilitated by the brilliant Richard Ranchurn, Senior Research Fellow and creative technologist at the Virtual Immersive Production (VIP) Studio at the University of Nottingham.
The cohort invited to the studio included artists, film makers and content producers. All of whom were able to spend the whole day exploring this fascinating piece of REAL-TIME AI video generation software.
Obviously, as you can see from the frenetic footage, itâs at a very early stage in its image generation (when you compare it to Kling or Nano-banana etc) - but it is also making 30 fps second - on a LOCAL COMPUTER! - Compared to 1 minute for 1 Adobe firefly image created on an army of data centres.
If nothing else, it felt like a new concept of live AI integration prompted by performers that felt controlled, and showing at least one way to control energy and data usage.
There are also layers to the prompting and scale of AI integration within this AI lens system that means you can scale up or down how much the AI imagery is integrated into your live filmed performance.
